[cdec.sa] Fix the list of matching training source sentence

3 files changed, 1455 insertions, 1389 deletions

diff --git a/python/src/sa/_sa.c b/python/src/sa/_sa.c
index 68205b2e..d1cc55c0 100644
--- a/python/src/sa/_sa.c
+++ b/python/src/sa/_sa.c
@@ -1,4 +1,4 @@
-/* Generated by Cython 0.17 on Thu Sep  6 11:14:53 2012 */
+/* Generated by Cython 0.17 on Thu Sep  6 14:37:00 2012 */
 
 #define PY_SSIZE_T_CLEAN
 #include "Python.h"
@@ -837,7 +837,7 @@ struct __pyx_obj_3_sa___pyx_scope_struct_6_genexpr {
 };
 
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":76
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":73
  *     def read_bitext(self, char* filename, int side):
  *         with gzip_or_text(filename) as fp:
  *             data = (line.split(' ||| ')[side] for line in fp)             # <<<<<<<<<<<<<<
@@ -889,7 +889,7 @@ struct __pyx_obj_3_sa___pyx_scope_struct_16___iter__ {
 };
 
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":74
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":71
  *             self.read_text_data(fp)
  * 
  *     def read_bitext(self, char* filename, int side):             # <<<<<<<<<<<<<<
@@ -2304,21 +2304,23 @@ static int __pyx_pf_3_sa_9StringMap___cinit__(struct __pyx_obj_3_sa_StringMap *_
 static void __pyx_pf_3_sa_9StringMap_2__dealloc__(struct __pyx_obj_3_sa_StringMap *__pyx_v_self); /* proto */
 static int __pyx_pf_3_sa_9DataArray___cinit__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_from_binary, PyObject *__pyx_v_from_text, PyObject *__pyx_v_side, int __pyx_v_use_sent_id); /* proto */
 static Py_ssize_t __pyx_pf_3_sa_9DataArray_2__len__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self); /* proto */
-static PyObject *__pyx_pf_3_sa_9DataArray_4get_data(struct __pyx_obj_3_sa_DataArray *__pyx_v_self); /* proto */
-static PyObject *__pyx_pf_3_sa_9DataArray_6get_sentence_id(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_i); /* proto */
-static PyObject *__pyx_pf_3_sa_9DataArray_8get_sentence(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_i); /* proto */
-static PyObject *__pyx_pf_3_sa_9DataArray_10get_sentence_position(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_loc); /* proto */
-static PyObject *__pyx_pf_3_sa_9DataArray_12get_id(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_word); /* proto */
-static PyObject *__pyx_pf_3_sa_9DataArray_14get_word(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_id); /* proto */
-static PyObject *__pyx_pf_3_sa_9DataArray_16write_text(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename); /* proto */
-static PyObject *__pyx_pf_3_sa_9DataArray_18read_text(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename); /* proto */
+static PyObject *__pyx_pf_3_sa_9DataArray_4get_sentence_id(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_i); /* proto */
+static PyObject *__pyx_pf_3_sa_9DataArray_6get_sentence(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_i); /* proto */
+static PyObject *__pyx_pf_3_sa_9DataArray_8get_sentence_position(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_loc); /* proto */
+static PyObject *__pyx_pf_3_sa_9DataArray_10get_id(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_word); /* proto */
+static PyObject *__pyx_pf_3_sa_9DataArray_12get_word(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_id); /* proto */
+static PyObject *__pyx_pf_3_sa_9DataArray_14write_text(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename); /* proto */
+static PyObject *__pyx_pf_3_sa_9DataArray_16read_text(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename); /* proto */
 static PyObject *__pyx_pf_3_sa_9DataArray_11read_bitext_genexpr(PyObject *__pyx_self); /* proto */
-static PyObject *__pyx_pf_3_sa_9DataArray_20read_bitext(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename, int __pyx_v_side); /* proto */
-static PyObject *__pyx_pf_3_sa_9DataArray_22read_text_data(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_data); /* proto */
-static PyObject *__pyx_pf_3_sa_9DataArray_24read_binary(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename); /* proto */
-static PyObject *__pyx_pf_3_sa_9DataArray_26write_binary(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename); /* proto */
-static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_f); /* proto */
-static PyObject *__pyx_pf_3_sa_9DataArray_30write_enhanced(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename); /* proto */
+static PyObject *__pyx_pf_3_sa_9DataArray_18read_bitext(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename, int __pyx_v_side); /* proto */
+static PyObject *__pyx_pf_3_sa_9DataArray_20read_text_data(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_data); /* proto */
+static PyObject *__pyx_pf_3_sa_9DataArray_22read_binary(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename); /* proto */
+static PyObject *__pyx_pf_3_sa_9DataArray_24write_binary(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename); /* proto */
+static PyObject *__pyx_pf_3_sa_9DataArray_26write_enhanced_handle(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_f); /* proto */
+static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename); /* proto */
+static PyObject *__pyx_pf_3_sa_9DataArray_4data___get__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self); /* proto */
+static int __pyx_pf_3_sa_9DataArray_4data_2__set__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_value); /* proto */
+static int __pyx_pf_3_sa_9DataArray_4data_4__del__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self); /* proto */
 static PyObject *__pyx_pf_3_sa_9Alignment_unlink(CYTHON_UNUSED struct __pyx_obj_3_sa_Alignment *__pyx_v_self, PyObject *__pyx_v_link); /* proto */
 static PyObject *__pyx_pf_3_sa_9Alignment_2get_sent_links(struct __pyx_obj_3_sa_Alignment *__pyx_v_self, int __pyx_v_sent_id); /* proto */
 static int __pyx_pf_3_sa_9Alignment_4__cinit__(struct __pyx_obj_3_sa_Alignment *__pyx_v_self, PyObject *__pyx_v_from_binary, PyObject *__pyx_v_from_text); /* proto */
@@ -2725,6 +2727,7 @@ static char __pyx_k__make_lattice[] = "make_lattice";
 static char __pyx_k__min_gap_size[] = "min_gap_size";
 static char __pyx_k__StopIteration[] = "StopIteration";
 static char __pyx_k__alphabet_size[] = "alphabet_size";
+static char __pyx_k__from_iterable[] = "from_iterable";
 static char __pyx_k__fsample_count[] = "fsample_count";
 static char __pyx_k__test_sentence[] = "test_sentence";
 static char __pyx_k__tight_phrases[] = "tight_phrases";
@@ -2900,6 +2903,7 @@ static PyObject *__pyx_n_s__filename;
 static PyObject *__pyx_n_s__fphrase;
 static PyObject *__pyx_n_s__from_binary;
 static PyObject *__pyx_n_s__from_data;
+static PyObject *__pyx_n_s__from_iterable;
 static PyObject *__pyx_n_s__from_stats;
 static PyObject *__pyx_n_s__from_text;
 static PyObject *__pyx_n_s__frontier;
@@ -7185,7 +7189,7 @@ static Py_ssize_t __pyx_pf_3_sa_9DataArray_2__len__(struct __pyx_obj_3_sa_DataAr
  *     def __len__(self):
  *         return len(self.data)             # <<<<<<<<<<<<<<
  * 
- *     def get_data(self):
+ *     def get_sentence_id(self, i):
  */
   __pyx_t_1 = ((PyObject *)__pyx_v_self->data);
   __Pyx_INCREF(__pyx_t_1);
@@ -7206,12 +7210,12 @@ static Py_ssize_t __pyx_pf_3_sa_9DataArray_2__len__(struct __pyx_obj_3_sa_DataAr
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_9DataArray_5get_data(PyObject *__pyx_v_self, CYTHON_UNUSED PyObject *unused); /*proto*/
-static PyObject *__pyx_pw_3_sa_9DataArray_5get_data(PyObject *__pyx_v_self, CYTHON_UNUSED PyObject *unused) {
+static PyObject *__pyx_pw_3_sa_9DataArray_5get_sentence_id(PyObject *__pyx_v_self, PyObject *__pyx_v_i); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_5get_sentence_id(PyObject *__pyx_v_self, PyObject *__pyx_v_i) {
   PyObject *__pyx_r = 0;
   __Pyx_RefNannyDeclarations
-  __Pyx_RefNannySetupContext("get_data (wrapper)", 0);
-  __pyx_r = __pyx_pf_3_sa_9DataArray_4get_data(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self));
+  __Pyx_RefNannySetupContext("get_sentence_id (wrapper)", 0);
+  __pyx_r = __pyx_pf_3_sa_9DataArray_4get_sentence_id(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((PyObject *)__pyx_v_i));
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
@@ -7219,55 +7223,12 @@ static PyObject *__pyx_pw_3_sa_9DataArray_5get_data(PyObject *__pyx_v_self, CYTH
 /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":35
  *         return len(self.data)
  * 
- *     def get_data(self):             # <<<<<<<<<<<<<<
- *         return self.data
- * 
- */
-
-static PyObject *__pyx_pf_3_sa_9DataArray_4get_data(struct __pyx_obj_3_sa_DataArray *__pyx_v_self) {
-  PyObject *__pyx_r = NULL;
-  __Pyx_RefNannyDeclarations
-  __Pyx_RefNannySetupContext("get_data", 0);
-
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":36
- * 
- *     def get_data(self):
- *         return self.data             # <<<<<<<<<<<<<<
- * 
- *     def get_sentence_id(self, i):
- */
-  __Pyx_XDECREF(__pyx_r);
-  __Pyx_INCREF(((PyObject *)__pyx_v_self->data));
-  __pyx_r = ((PyObject *)__pyx_v_self->data);
-  goto __pyx_L0;
-
-  __pyx_r = Py_None; __Pyx_INCREF(Py_None);
-  __pyx_L0:;
-  __Pyx_XGIVEREF(__pyx_r);
-  __Pyx_RefNannyFinishContext();
-  return __pyx_r;
-}
-
-/* Python wrapper */
-static PyObject *__pyx_pw_3_sa_9DataArray_7get_sentence_id(PyObject *__pyx_v_self, PyObject *__pyx_v_i); /*proto*/
-static PyObject *__pyx_pw_3_sa_9DataArray_7get_sentence_id(PyObject *__pyx_v_self, PyObject *__pyx_v_i) {
-  PyObject *__pyx_r = 0;
-  __Pyx_RefNannyDeclarations
-  __Pyx_RefNannySetupContext("get_sentence_id (wrapper)", 0);
-  __pyx_r = __pyx_pf_3_sa_9DataArray_6get_sentence_id(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((PyObject *)__pyx_v_i));
-  __Pyx_RefNannyFinishContext();
-  return __pyx_r;
-}
-
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":38
- *         return self.data
- * 
  *     def get_sentence_id(self, i):             # <<<<<<<<<<<<<<
  *         return self.sent_id.arr[i]
  * 
  */
 
-static PyObject *__pyx_pf_3_sa_9DataArray_6get_sentence_id(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_i) {
+static PyObject *__pyx_pf_3_sa_9DataArray_4get_sentence_id(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_i) {
   PyObject *__pyx_r = NULL;
   __Pyx_RefNannyDeclarations
   Py_ssize_t __pyx_t_1;
@@ -7277,7 +7238,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_6get_sentence_id(struct __pyx_obj_3_sa
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("get_sentence_id", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":39
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":36
  * 
  *     def get_sentence_id(self, i):
  *         return self.sent_id.arr[i]             # <<<<<<<<<<<<<<
@@ -7285,8 +7246,8 @@ static PyObject *__pyx_pf_3_sa_9DataArray_6get_sentence_id(struct __pyx_obj_3_sa
  *     def get_sentence(self, i):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_1 = __Pyx_PyIndex_AsSsize_t(__pyx_v_i); if (unlikely((__pyx_t_1 == (Py_ssize_t)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 39; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __pyx_t_2 = PyInt_FromLong((__pyx_v_self->sent_id->arr[__pyx_t_1])); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 39; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyIndex_AsSsize_t(__pyx_v_i); if (unlikely((__pyx_t_1 == (Py_ssize_t)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 36; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyInt_FromLong((__pyx_v_self->sent_id->arr[__pyx_t_1])); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 36; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -7305,17 +7266,17 @@ static PyObject *__pyx_pf_3_sa_9DataArray_6get_sentence_id(struct __pyx_obj_3_sa
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_9DataArray_9get_sentence(PyObject *__pyx_v_self, PyObject *__pyx_v_i); /*proto*/
-static PyObject *__pyx_pw_3_sa_9DataArray_9get_sentence(PyObject *__pyx_v_self, PyObject *__pyx_v_i) {
+static PyObject *__pyx_pw_3_sa_9DataArray_7get_sentence(PyObject *__pyx_v_self, PyObject *__pyx_v_i); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_7get_sentence(PyObject *__pyx_v_self, PyObject *__pyx_v_i) {
   PyObject *__pyx_r = 0;
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("get_sentence (wrapper)", 0);
-  __pyx_r = __pyx_pf_3_sa_9DataArray_8get_sentence(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((PyObject *)__pyx_v_i));
+  __pyx_r = __pyx_pf_3_sa_9DataArray_6get_sentence(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((PyObject *)__pyx_v_i));
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":41
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":38
  *         return self.sent_id.arr[i]
  * 
  *     def get_sentence(self, i):             # <<<<<<<<<<<<<<
@@ -7323,7 +7284,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_9get_sentence(PyObject *__pyx_v_self,
  *         sent = []
  */
 
-static PyObject *__pyx_pf_3_sa_9DataArray_8get_sentence(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_i) {
+static PyObject *__pyx_pf_3_sa_9DataArray_6get_sentence(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_i) {
   int __pyx_v_start;
   int __pyx_v_stop;
   PyObject *__pyx_v_sent = NULL;
@@ -7340,42 +7301,42 @@ static PyObject *__pyx_pf_3_sa_9DataArray_8get_sentence(struct __pyx_obj_3_sa_Da
   __Pyx_RefNannySetupContext("get_sentence", 0);
   __Pyx_INCREF(__pyx_v_i);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":43
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":40
  *     def get_sentence(self, i):
  *         cdef int j, start, stop
  *         sent = []             # <<<<<<<<<<<<<<
  *         start = self.sent_index.arr[i]
  *         stop = self.sent_index.arr[i+1]
  */
-  __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 43; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 40; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __pyx_v_sent = __pyx_t_1;
   __pyx_t_1 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":44
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":41
  *         cdef int j, start, stop
  *         sent = []
  *         start = self.sent_index.arr[i]             # <<<<<<<<<<<<<<
  *         stop = self.sent_index.arr[i+1]
  *         for i from start <= i < stop:
  */
-  __pyx_t_2 = __Pyx_PyIndex_AsSsize_t(__pyx_v_i); if (unlikely((__pyx_t_2 == (Py_ssize_t)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 44; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_PyIndex_AsSsize_t(__pyx_v_i); if (unlikely((__pyx_t_2 == (Py_ssize_t)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 41; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_v_start = (__pyx_v_self->sent_index->arr[__pyx_t_2]);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":45
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":42
  *         sent = []
  *         start = self.sent_index.arr[i]
  *         stop = self.sent_index.arr[i+1]             # <<<<<<<<<<<<<<
  *         for i from start <= i < stop:
  *             sent.append(self.id2word[self.data.arr[i]])
  */
-  __pyx_t_1 = PyNumber_Add(__pyx_v_i, __pyx_int_1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 45; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyNumber_Add(__pyx_v_i, __pyx_int_1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 42; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_2 = __Pyx_PyIndex_AsSsize_t(__pyx_t_1); if (unlikely((__pyx_t_2 == (Py_ssize_t)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 45; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_PyIndex_AsSsize_t(__pyx_t_1); if (unlikely((__pyx_t_2 == (Py_ssize_t)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 42; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
   __pyx_v_stop = (__pyx_v_self->sent_index->arr[__pyx_t_2]);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":46
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":43
  *         start = self.sent_index.arr[i]
  *         stop = self.sent_index.arr[i+1]
  *         for i from start <= i < stop:             # <<<<<<<<<<<<<<
@@ -7384,41 +7345,41 @@ static PyObject *__pyx_pf_3_sa_9DataArray_8get_sentence(struct __pyx_obj_3_sa_Da
  */
   __pyx_t_3 = __pyx_v_stop;
   for (__pyx_t_4 = __pyx_v_start; __pyx_t_4 < __pyx_t_3; __pyx_t_4++) {
-    __pyx_t_1 = PyInt_FromLong(__pyx_t_4); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 46; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = PyInt_FromLong(__pyx_t_4); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 43; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
     __Pyx_DECREF(__pyx_v_i);
     __pyx_v_i = __pyx_t_1;
     __pyx_t_1 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":47
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":44
  *         stop = self.sent_index.arr[i+1]
  *         for i from start <= i < stop:
  *             sent.append(self.id2word[self.data.arr[i]])             # <<<<<<<<<<<<<<
  *         return sent
  * 
  */
-    __pyx_t_2 = __Pyx_PyIndex_AsSsize_t(__pyx_v_i); if (unlikely((__pyx_t_2 == (Py_ssize_t)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 47; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-    __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_self->id2word, (__pyx_v_self->data->arr[__pyx_t_2]), sizeof(int), PyInt_FromLong); if (!__pyx_t_1) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 47; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __Pyx_PyIndex_AsSsize_t(__pyx_v_i); if (unlikely((__pyx_t_2 == (Py_ssize_t)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 44; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_self->id2word, (__pyx_v_self->data->arr[__pyx_t_2]), sizeof(int), PyInt_FromLong); if (!__pyx_t_1) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 44; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
-    __pyx_t_5 = PyList_Append(__pyx_v_sent, __pyx_t_1); if (unlikely(__pyx_t_5 == -1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 47; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyList_Append(__pyx_v_sent, __pyx_t_1); if (unlikely(__pyx_t_5 == -1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 44; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-    __pyx_t_4 = __Pyx_PyInt_AsInt(__pyx_v_i); if (unlikely((__pyx_t_4 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 46; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = __Pyx_PyInt_AsInt(__pyx_v_i); if (unlikely((__pyx_t_4 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 43; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":46
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":43
  *         start = self.sent_index.arr[i]
  *         stop = self.sent_index.arr[i+1]
  *         for i from start <= i < stop:             # <<<<<<<<<<<<<<
  *             sent.append(self.id2word[self.data.arr[i]])
  *         return sent
  */
-  __pyx_t_1 = PyInt_FromLong(__pyx_t_4); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 46; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyInt_FromLong(__pyx_t_4); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 43; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_v_i);
   __pyx_v_i = __pyx_t_1;
   __pyx_t_1 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":48
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":45
  *         for i from start <= i < stop:
  *             sent.append(self.id2word[self.data.arr[i]])
  *         return sent             # <<<<<<<<<<<<<<
@@ -7445,17 +7406,17 @@ static PyObject *__pyx_pf_3_sa_9DataArray_8get_sentence(struct __pyx_obj_3_sa_Da
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_9DataArray_11get_sentence_position(PyObject *__pyx_v_self, PyObject *__pyx_v_loc); /*proto*/
-static PyObject *__pyx_pw_3_sa_9DataArray_11get_sentence_position(PyObject *__pyx_v_self, PyObject *__pyx_v_loc) {
+static PyObject *__pyx_pw_3_sa_9DataArray_9get_sentence_position(PyObject *__pyx_v_self, PyObject *__pyx_v_loc); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_9get_sentence_position(PyObject *__pyx_v_self, PyObject *__pyx_v_loc) {
   PyObject *__pyx_r = 0;
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("get_sentence_position (wrapper)", 0);
-  __pyx_r = __pyx_pf_3_sa_9DataArray_10get_sentence_position(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((PyObject *)__pyx_v_loc));
+  __pyx_r = __pyx_pf_3_sa_9DataArray_8get_sentence_position(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((PyObject *)__pyx_v_loc));
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":50
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":47
  *         return sent
  * 
  *     def get_sentence_position(self, loc):             # <<<<<<<<<<<<<<
@@ -7463,7 +7424,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_11get_sentence_position(PyObject *__py
  * 
  */
 
-static PyObject *__pyx_pf_3_sa_9DataArray_10get_sentence_position(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_loc) {
+static PyObject *__pyx_pf_3_sa_9DataArray_8get_sentence_position(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_loc) {
   PyObject *__pyx_r = NULL;
   __Pyx_RefNannyDeclarations
   Py_ssize_t __pyx_t_1;
@@ -7474,7 +7435,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_10get_sentence_position(struct __pyx_o
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("get_sentence_position", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":51
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":48
  * 
  *     def get_sentence_position(self, loc):
  *         return loc - self.sent_index.arr[self.sent_id.arr[loc]]             # <<<<<<<<<<<<<<
@@ -7482,10 +7443,10 @@ static PyObject *__pyx_pf_3_sa_9DataArray_10get_sentence_position(struct __pyx_o
  *     def get_id(self, word):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_1 = __Pyx_PyIndex_AsSsize_t(__pyx_v_loc); if (unlikely((__pyx_t_1 == (Py_ssize_t)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 51; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __pyx_t_2 = PyInt_FromLong((__pyx_v_self->sent_index->arr[(__pyx_v_self->sent_id->arr[__pyx_t_1])])); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 51; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyIndex_AsSsize_t(__pyx_v_loc); if (unlikely((__pyx_t_1 == (Py_ssize_t)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 48; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyInt_FromLong((__pyx_v_self->sent_index->arr[(__pyx_v_self->sent_id->arr[__pyx_t_1])])); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 48; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyNumber_Subtract(__pyx_v_loc, __pyx_t_2); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 51; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyNumber_Subtract(__pyx_v_loc, __pyx_t_2); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 48; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __pyx_r = __pyx_t_3;
@@ -7506,17 +7467,17 @@ static PyObject *__pyx_pf_3_sa_9DataArray_10get_sentence_position(struct __pyx_o
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_9DataArray_13get_id(PyObject *__pyx_v_self, PyObject *__pyx_v_word); /*proto*/
-static PyObject *__pyx_pw_3_sa_9DataArray_13get_id(PyObject *__pyx_v_self, PyObject *__pyx_v_word) {
+static PyObject *__pyx_pw_3_sa_9DataArray_11get_id(PyObject *__pyx_v_self, PyObject *__pyx_v_word); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_11get_id(PyObject *__pyx_v_self, PyObject *__pyx_v_word) {
   PyObject *__pyx_r = 0;
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("get_id (wrapper)", 0);
-  __pyx_r = __pyx_pf_3_sa_9DataArray_12get_id(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((PyObject *)__pyx_v_word));
+  __pyx_r = __pyx_pf_3_sa_9DataArray_10get_id(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((PyObject *)__pyx_v_word));
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":53
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":50
  *         return loc - self.sent_index.arr[self.sent_id.arr[loc]]
  * 
  *     def get_id(self, word):             # <<<<<<<<<<<<<<
@@ -7524,7 +7485,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_13get_id(PyObject *__pyx_v_self, PyObj
  *             self.word2id[word] = len(self.id2word)
  */
 
-static PyObject *__pyx_pf_3_sa_9DataArray_12get_id(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_word) {
+static PyObject *__pyx_pf_3_sa_9DataArray_10get_id(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_word) {
   PyObject *__pyx_r = NULL;
   __Pyx_RefNannyDeclarations
   int __pyx_t_1;
@@ -7536,18 +7497,18 @@ static PyObject *__pyx_pf_3_sa_9DataArray_12get_id(struct __pyx_obj_3_sa_DataArr
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("get_id", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":54
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":51
  * 
  *     def get_id(self, word):
  *         if not word in self.word2id:             # <<<<<<<<<<<<<<
  *             self.word2id[word] = len(self.id2word)
  *             self.id2word.append(word)
  */
-  __pyx_t_1 = (__Pyx_PySequence_Contains(__pyx_v_word, __pyx_v_self->word2id, Py_EQ)); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 54; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = (__Pyx_PySequence_Contains(__pyx_v_word, __pyx_v_self->word2id, Py_EQ)); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 51; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_t_2 = (!__pyx_t_1);
   if (__pyx_t_2) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":55
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":52
  *     def get_id(self, word):
  *         if not word in self.word2id:
  *             self.word2id[word] = len(self.id2word)             # <<<<<<<<<<<<<<
@@ -7556,28 +7517,28 @@ static PyObject *__pyx_pf_3_sa_9DataArray_12get_id(struct __pyx_obj_3_sa_DataArr
  */
     __pyx_t_3 = __pyx_v_self->id2word;
     __Pyx_INCREF(__pyx_t_3);
-    __pyx_t_4 = PyObject_Length(__pyx_t_3); if (unlikely(__pyx_t_4 == -1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 55; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyObject_Length(__pyx_t_3); if (unlikely(__pyx_t_4 == -1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 52; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-    __pyx_t_3 = PyInt_FromSsize_t(__pyx_t_4); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 55; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyInt_FromSsize_t(__pyx_t_4); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 52; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
-    if (PyObject_SetItem(__pyx_v_self->word2id, __pyx_v_word, __pyx_t_3) < 0) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 55; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    if (PyObject_SetItem(__pyx_v_self->word2id, __pyx_v_word, __pyx_t_3) < 0) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 52; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":56
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":53
  *         if not word in self.word2id:
  *             self.word2id[word] = len(self.id2word)
  *             self.id2word.append(word)             # <<<<<<<<<<<<<<
  *         return self.word2id[word]
  * 
  */
-    __pyx_t_3 = __Pyx_PyObject_Append(__pyx_v_self->id2word, __pyx_v_word); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 56; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = __Pyx_PyObject_Append(__pyx_v_self->id2word, __pyx_v_word); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 53; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     goto __pyx_L3;
   }
   __pyx_L3:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":57
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":54
  *             self.word2id[word] = len(self.id2word)
  *             self.id2word.append(word)
  *         return self.word2id[word]             # <<<<<<<<<<<<<<
@@ -7585,7 +7546,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_12get_id(struct __pyx_obj_3_sa_DataArr
  *     def get_word(self, id):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_3 = PyObject_GetItem(__pyx_v_self->word2id, __pyx_v_word); if (!__pyx_t_3) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 57; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetItem(__pyx_v_self->word2id, __pyx_v_word); if (!__pyx_t_3) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 54; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __pyx_r = __pyx_t_3;
   __pyx_t_3 = 0;
@@ -7604,17 +7565,17 @@ static PyObject *__pyx_pf_3_sa_9DataArray_12get_id(struct __pyx_obj_3_sa_DataArr
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_9DataArray_15get_word(PyObject *__pyx_v_self, PyObject *__pyx_v_id); /*proto*/
-static PyObject *__pyx_pw_3_sa_9DataArray_15get_word(PyObject *__pyx_v_self, PyObject *__pyx_v_id) {
+static PyObject *__pyx_pw_3_sa_9DataArray_13get_word(PyObject *__pyx_v_self, PyObject *__pyx_v_id); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_13get_word(PyObject *__pyx_v_self, PyObject *__pyx_v_id) {
   PyObject *__pyx_r = 0;
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("get_word (wrapper)", 0);
-  __pyx_r = __pyx_pf_3_sa_9DataArray_14get_word(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((PyObject *)__pyx_v_id));
+  __pyx_r = __pyx_pf_3_sa_9DataArray_12get_word(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((PyObject *)__pyx_v_id));
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":59
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":56
  *         return self.word2id[word]
  * 
  *     def get_word(self, id):             # <<<<<<<<<<<<<<
@@ -7622,7 +7583,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_15get_word(PyObject *__pyx_v_self, PyO
  * 
  */
 
-static PyObject *__pyx_pf_3_sa_9DataArray_14get_word(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_id) {
+static PyObject *__pyx_pf_3_sa_9DataArray_12get_word(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_id) {
   PyObject *__pyx_r = NULL;
   __Pyx_RefNannyDeclarations
   PyObject *__pyx_t_1 = NULL;
@@ -7631,7 +7592,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_14get_word(struct __pyx_obj_3_sa_DataA
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("get_word", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":60
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":57
  * 
  *     def get_word(self, id):
  *         return self.id2word[id]             # <<<<<<<<<<<<<<
@@ -7639,7 +7600,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_14get_word(struct __pyx_obj_3_sa_DataA
  *     def write_text(self, char* filename):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_1 = PyObject_GetItem(__pyx_v_self->id2word, __pyx_v_id); if (!__pyx_t_1) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 60; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetItem(__pyx_v_self->id2word, __pyx_v_id); if (!__pyx_t_1) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 57; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __pyx_r = __pyx_t_1;
   __pyx_t_1 = 0;
@@ -7658,14 +7619,14 @@ static PyObject *__pyx_pf_3_sa_9DataArray_14get_word(struct __pyx_obj_3_sa_DataA
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_9DataArray_17write_text(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename); /*proto*/
-static PyObject *__pyx_pw_3_sa_9DataArray_17write_text(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename) {
+static PyObject *__pyx_pw_3_sa_9DataArray_15write_text(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_15write_text(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename) {
   char *__pyx_v_filename;
   PyObject *__pyx_r = 0;
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("write_text (wrapper)", 0);
   assert(__pyx_arg_filename); {
-    __pyx_v_filename = PyBytes_AsString(__pyx_arg_filename); if (unlikely((!__pyx_v_filename) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 62; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_v_filename = PyBytes_AsString(__pyx_arg_filename); if (unlikely((!__pyx_v_filename) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 59; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L3_error:;
@@ -7673,12 +7634,12 @@ static PyObject *__pyx_pw_3_sa_9DataArray_17write_text(PyObject *__pyx_v_self, P
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_r = __pyx_pf_3_sa_9DataArray_16write_text(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((char *)__pyx_v_filename));
+  __pyx_r = __pyx_pf_3_sa_9DataArray_14write_text(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((char *)__pyx_v_filename));
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":62
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":59
  *         return self.id2word[id]
  * 
  *     def write_text(self, char* filename):             # <<<<<<<<<<<<<<
@@ -7686,7 +7647,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_17write_text(PyObject *__pyx_v_self, P
  *             for w_id in self.data:
  */
 
-static PyObject *__pyx_pf_3_sa_9DataArray_16write_text(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename) {
+static PyObject *__pyx_pf_3_sa_9DataArray_14write_text(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename) {
   PyObject *__pyx_v_f = NULL;
   PyObject *__pyx_v_w_id = NULL;
   PyObject *__pyx_r = NULL;
@@ -7710,7 +7671,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16write_text(struct __pyx_obj_3_sa_Dat
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("write_text", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":63
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":60
  * 
  *     def write_text(self, char* filename):
  *         with open(filename, "w") as f:             # <<<<<<<<<<<<<<
@@ -7718,9 +7679,9 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16write_text(struct __pyx_obj_3_sa_Dat
  *                 if w_id > 1:
  */
   /*with:*/ {
-    __pyx_t_1 = PyBytes_FromString(__pyx_v_filename); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = PyBytes_FromString(__pyx_v_filename); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 60; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_1));
-    __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 60; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_t_1));
     __Pyx_GIVEREF(((PyObject *)__pyx_t_1));
@@ -7728,14 +7689,14 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16write_text(struct __pyx_obj_3_sa_Dat
     PyTuple_SET_ITEM(__pyx_t_2, 1, ((PyObject *)__pyx_n_s__w));
     __Pyx_GIVEREF(((PyObject *)__pyx_n_s__w));
     __pyx_t_1 = 0;
-    __pyx_t_1 = PyObject_Call(__pyx_builtin_open, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = PyObject_Call(__pyx_builtin_open, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 60; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
     __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
-    __pyx_t_3 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s____exit__); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s____exit__); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 60; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
-    __pyx_t_2 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s____enter__); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_t_2 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s____enter__); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 60; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_4 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_t_4 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 60; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     __Pyx_GOTREF(__pyx_t_4);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
     __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
@@ -7750,7 +7711,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16write_text(struct __pyx_obj_3_sa_Dat
           __pyx_v_f = __pyx_t_4;
           __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":64
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":61
  *     def write_text(self, char* filename):
  *         with open(filename, "w") as f:
  *             for w_id in self.data:             # <<<<<<<<<<<<<<
@@ -7761,7 +7722,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16write_text(struct __pyx_obj_3_sa_Dat
             __pyx_t_4 = ((PyObject *)__pyx_v_self->data); __Pyx_INCREF(__pyx_t_4); __pyx_t_8 = 0;
             __pyx_t_9 = NULL;
           } else {
-            __pyx_t_8 = -1; __pyx_t_4 = PyObject_GetIter(((PyObject *)__pyx_v_self->data)); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 64; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+            __pyx_t_8 = -1; __pyx_t_4 = PyObject_GetIter(((PyObject *)__pyx_v_self->data)); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 61; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
             __Pyx_GOTREF(__pyx_t_4);
             __pyx_t_9 = Py_TYPE(__pyx_t_4)->tp_iternext;
           }
@@ -7769,23 +7730,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16write_text(struct __pyx_obj_3_sa_Dat
             if (!__pyx_t_9 && PyList_CheckExact(__pyx_t_4)) {
               if (__pyx_t_8 >= PyList_GET_SIZE(__pyx_t_4)) break;
               #if CYTHON_COMPILING_IN_CPYTHON
-              __pyx_t_1 = PyList_GET_ITEM(__pyx_t_4, __pyx_t_8); __Pyx_INCREF(__pyx_t_1); __pyx_t_8++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 64; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_1 = PyList_GET_ITEM(__pyx_t_4, __pyx_t_8); __Pyx_INCREF(__pyx_t_1); __pyx_t_8++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 61; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               #else
-              __pyx_t_1 = PySequence_ITEM(__pyx_t_4, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 64; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_1 = PySequence_ITEM(__pyx_t_4, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 61; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               #endif
             } else if (!__pyx_t_9 && PyTuple_CheckExact(__pyx_t_4)) {
               if (__pyx_t_8 >= PyTuple_GET_SIZE(__pyx_t_4)) break;
               #if CYTHON_COMPILING_IN_CPYTHON
-              __pyx_t_1 = PyTuple_GET_ITEM(__pyx_t_4, __pyx_t_8); __Pyx_INCREF(__pyx_t_1); __pyx_t_8++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 64; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_1 = PyTuple_GET_ITEM(__pyx_t_4, __pyx_t_8); __Pyx_INCREF(__pyx_t_1); __pyx_t_8++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 61; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               #else
-              __pyx_t_1 = PySequence_ITEM(__pyx_t_4, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 64; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_1 = PySequence_ITEM(__pyx_t_4, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 61; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               #endif
             } else {
               __pyx_t_1 = __pyx_t_9(__pyx_t_4);
               if (unlikely(!__pyx_t_1)) {
                 if (PyErr_Occurred()) {
                   if (likely(PyErr_ExceptionMatches(PyExc_StopIteration))) PyErr_Clear();
-                  else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 64; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+                  else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 61; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
                 }
                 break;
               }
@@ -7795,47 +7756,47 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16write_text(struct __pyx_obj_3_sa_Dat
             __pyx_v_w_id = __pyx_t_1;
             __pyx_t_1 = 0;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":65
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":62
  *         with open(filename, "w") as f:
  *             for w_id in self.data:
  *                 if w_id > 1:             # <<<<<<<<<<<<<<
  *                     f.write("%s " % self.get_word(w_id))
  *                 if w_id == 1:
  */
-            __pyx_t_1 = PyObject_RichCompare(__pyx_v_w_id, __pyx_int_1, Py_GT); __Pyx_XGOTREF(__pyx_t_1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 65; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
-            __pyx_t_10 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_10 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 65; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+            __pyx_t_1 = PyObject_RichCompare(__pyx_v_w_id, __pyx_int_1, Py_GT); __Pyx_XGOTREF(__pyx_t_1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 62; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+            __pyx_t_10 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_10 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 62; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
             __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
             if (__pyx_t_10) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":66
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":63
  *             for w_id in self.data:
  *                 if w_id > 1:
  *                     f.write("%s " % self.get_word(w_id))             # <<<<<<<<<<<<<<
  *                 if w_id == 1:
  *                     f.write("\n")
  */
-              __pyx_t_1 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 66; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_1 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               __Pyx_GOTREF(__pyx_t_1);
-              __pyx_t_2 = PyObject_GetAttr(((PyObject *)__pyx_v_self), __pyx_n_s__get_word); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 66; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_2 = PyObject_GetAttr(((PyObject *)__pyx_v_self), __pyx_n_s__get_word); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               __Pyx_GOTREF(__pyx_t_2);
-              __pyx_t_11 = PyTuple_New(1); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 66; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_11 = PyTuple_New(1); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               __Pyx_GOTREF(__pyx_t_11);
               __Pyx_INCREF(__pyx_v_w_id);
               PyTuple_SET_ITEM(__pyx_t_11, 0, __pyx_v_w_id);
               __Pyx_GIVEREF(__pyx_v_w_id);
-              __pyx_t_12 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_11), NULL); if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 66; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_12 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_11), NULL); if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               __Pyx_GOTREF(__pyx_t_12);
               __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
               __Pyx_DECREF(((PyObject *)__pyx_t_11)); __pyx_t_11 = 0;
-              __pyx_t_11 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_13), __pyx_t_12); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 66; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_11 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_13), __pyx_t_12); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               __Pyx_GOTREF(((PyObject *)__pyx_t_11));
               __Pyx_DECREF(__pyx_t_12); __pyx_t_12 = 0;
-              __pyx_t_12 = PyTuple_New(1); if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 66; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_12 = PyTuple_New(1); if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               __Pyx_GOTREF(__pyx_t_12);
               PyTuple_SET_ITEM(__pyx_t_12, 0, ((PyObject *)__pyx_t_11));
               __Pyx_GIVEREF(((PyObject *)__pyx_t_11));
               __pyx_t_11 = 0;
-              __pyx_t_11 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_12), NULL); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 66; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_11 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_12), NULL); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               __Pyx_GOTREF(__pyx_t_11);
               __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
               __Pyx_DECREF(((PyObject *)__pyx_t_12)); __pyx_t_12 = 0;
@@ -7844,28 +7805,28 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16write_text(struct __pyx_obj_3_sa_Dat
             }
             __pyx_L18:;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":67
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":64
  *                 if w_id > 1:
  *                     f.write("%s " % self.get_word(w_id))
  *                 if w_id == 1:             # <<<<<<<<<<<<<<
  *                     f.write("\n")
  * 
  */
-            __pyx_t_11 = PyObject_RichCompare(__pyx_v_w_id, __pyx_int_1, Py_EQ); __Pyx_XGOTREF(__pyx_t_11); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 67; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
-            __pyx_t_10 = __Pyx_PyObject_IsTrue(__pyx_t_11); if (unlikely(__pyx_t_10 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 67; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+            __pyx_t_11 = PyObject_RichCompare(__pyx_v_w_id, __pyx_int_1, Py_EQ); __Pyx_XGOTREF(__pyx_t_11); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 64; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+            __pyx_t_10 = __Pyx_PyObject_IsTrue(__pyx_t_11); if (unlikely(__pyx_t_10 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 64; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
             __Pyx_DECREF(__pyx_t_11); __pyx_t_11 = 0;
             if (__pyx_t_10) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":68
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":65
  *                     f.write("%s " % self.get_word(w_id))
  *                 if w_id == 1:
  *                     f.write("\n")             # <<<<<<<<<<<<<<
  * 
  *     def read_text(self, char* filename):
  */
-              __pyx_t_11 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_11 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 65; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               __Pyx_GOTREF(__pyx_t_11);
-              __pyx_t_12 = PyObject_Call(__pyx_t_11, ((PyObject *)__pyx_k_tuple_15), NULL); if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_12 = PyObject_Call(__pyx_t_11, ((PyObject *)__pyx_k_tuple_15), NULL); if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 65; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               __Pyx_GOTREF(__pyx_t_12);
               __Pyx_DECREF(__pyx_t_11); __pyx_t_11 = 0;
               __Pyx_DECREF(__pyx_t_12); __pyx_t_12 = 0;
@@ -7886,7 +7847,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16write_text(struct __pyx_obj_3_sa_Dat
         __Pyx_XDECREF(__pyx_t_12); __pyx_t_12 = 0;
         __Pyx_XDECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":63
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":60
  * 
  *     def write_text(self, char* filename):
  *         with open(filename, "w") as f:             # <<<<<<<<<<<<<<
@@ -7895,11 +7856,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16write_text(struct __pyx_obj_3_sa_Dat
  */
         /*except:*/ {
           __Pyx_AddTraceback("_sa.DataArray.write_text", __pyx_clineno, __pyx_lineno, __pyx_filename);
-          if (__Pyx_GetException(&__pyx_t_4, &__pyx_t_12, &__pyx_t_11) < 0) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (__Pyx_GetException(&__pyx_t_4, &__pyx_t_12, &__pyx_t_11) < 0) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 60; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __Pyx_GOTREF(__pyx_t_4);
           __Pyx_GOTREF(__pyx_t_12);
           __Pyx_GOTREF(__pyx_t_11);
-          __pyx_t_1 = PyTuple_New(3); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          __pyx_t_1 = PyTuple_New(3); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 60; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __Pyx_GOTREF(__pyx_t_1);
           __Pyx_INCREF(__pyx_t_4);
           PyTuple_SET_ITEM(__pyx_t_1, 0, __pyx_t_4);
@@ -7912,11 +7873,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16write_text(struct __pyx_obj_3_sa_Dat
           __Pyx_GIVEREF(__pyx_t_11);
           __pyx_t_13 = PyObject_Call(__pyx_t_3, __pyx_t_1, NULL);
           __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-          if (unlikely(!__pyx_t_13)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (unlikely(!__pyx_t_13)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 60; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __Pyx_GOTREF(__pyx_t_13);
           __pyx_t_10 = __Pyx_PyObject_IsTrue(__pyx_t_13);
           __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0;
-          if (unlikely(__pyx_t_10 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (unlikely(__pyx_t_10 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 60; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __pyx_t_14 = (!__pyx_t_10);
           if (__pyx_t_14) {
             __Pyx_GIVEREF(__pyx_t_4);
@@ -7924,7 +7885,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16write_text(struct __pyx_obj_3_sa_Dat
             __Pyx_GIVEREF(__pyx_t_11);
             __Pyx_ErrRestore(__pyx_t_4, __pyx_t_12, __pyx_t_11);
             __pyx_t_4 = 0; __pyx_t_12 = 0; __pyx_t_11 = 0; 
-            {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+            {__pyx_filename = __pyx_f[3]; __pyx_lineno = 60; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
             goto __pyx_L22;
           }
           __pyx_L22:;
@@ -7952,11 +7913,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16write_text(struct __pyx_obj_3_sa_Dat
       if (__pyx_t_3) {
         __pyx_t_7 = PyObject_Call(__pyx_t_3, __pyx_k_tuple_16, NULL);
         __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-        if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 60; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_7);
         __pyx_t_14 = __Pyx_PyObject_IsTrue(__pyx_t_7);
         __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
-        if (unlikely(__pyx_t_14 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        if (unlikely(__pyx_t_14 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 60; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       }
     }
     goto __pyx_L23;
@@ -7985,14 +7946,14 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16write_text(struct __pyx_obj_3_sa_Dat
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_9DataArray_19read_text(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename); /*proto*/
-static PyObject *__pyx_pw_3_sa_9DataArray_19read_text(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename) {
+static PyObject *__pyx_pw_3_sa_9DataArray_17read_text(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_17read_text(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename) {
   char *__pyx_v_filename;
   PyObject *__pyx_r = 0;
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("read_text (wrapper)", 0);
   assert(__pyx_arg_filename); {
-    __pyx_v_filename = PyBytes_AsString(__pyx_arg_filename); if (unlikely((!__pyx_v_filename) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 70; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_v_filename = PyBytes_AsString(__pyx_arg_filename); if (unlikely((!__pyx_v_filename) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 67; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L3_error:;
@@ -8000,12 +7961,12 @@ static PyObject *__pyx_pw_3_sa_9DataArray_19read_text(PyObject *__pyx_v_self, Py
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_r = __pyx_pf_3_sa_9DataArray_18read_text(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((char *)__pyx_v_filename));
+  __pyx_r = __pyx_pf_3_sa_9DataArray_16read_text(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((char *)__pyx_v_filename));
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":70
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":67
  *                     f.write("\n")
  * 
  *     def read_text(self, char* filename):             # <<<<<<<<<<<<<<
@@ -8013,7 +7974,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_19read_text(PyObject *__pyx_v_self, Py
  *             self.read_text_data(fp)
  */
 
-static PyObject *__pyx_pf_3_sa_9DataArray_18read_text(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename) {
+static PyObject *__pyx_pf_3_sa_9DataArray_16read_text(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename) {
   PyObject *__pyx_v_fp = NULL;
   PyObject *__pyx_r = NULL;
   __Pyx_RefNannyDeclarations
@@ -8033,7 +7994,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_18read_text(struct __pyx_obj_3_sa_Data
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("read_text", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":71
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":68
  * 
  *     def read_text(self, char* filename):
  *         with gzip_or_text(filename) as fp:             # <<<<<<<<<<<<<<
@@ -8041,24 +8002,24 @@ static PyObject *__pyx_pf_3_sa_9DataArray_18read_text(struct __pyx_obj_3_sa_Data
  * 
  */
   /*with:*/ {
-    __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s__gzip_or_text); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s__gzip_or_text); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
-    __pyx_t_2 = PyBytes_FromString(__pyx_v_filename); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyBytes_FromString(__pyx_v_filename); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_2));
-    __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
     PyTuple_SET_ITEM(__pyx_t_3, 0, ((PyObject *)__pyx_t_2));
     __Pyx_GIVEREF(((PyObject *)__pyx_t_2));
     __pyx_t_2 = 0;
-    __pyx_t_2 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
-    __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s____exit__); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s____exit__); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
-    __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s____enter__); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s____enter__); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     __Pyx_GOTREF(__pyx_t_3);
-    __pyx_t_1 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_t_1 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     __Pyx_GOTREF(__pyx_t_1);
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
@@ -8073,21 +8034,21 @@ static PyObject *__pyx_pf_3_sa_9DataArray_18read_text(struct __pyx_obj_3_sa_Data
           __pyx_v_fp = __pyx_t_1;
           __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":72
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":69
  *     def read_text(self, char* filename):
  *         with gzip_or_text(filename) as fp:
  *             self.read_text_data(fp)             # <<<<<<<<<<<<<<
  * 
  *     def read_bitext(self, char* filename, int side):
  */
-          __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self), __pyx_n_s__read_text_data); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+          __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self), __pyx_n_s__read_text_data); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 69; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
           __Pyx_GOTREF(__pyx_t_1);
-          __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+          __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 69; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
           __Pyx_GOTREF(__pyx_t_2);
           __Pyx_INCREF(__pyx_v_fp);
           PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_v_fp);
           __Pyx_GIVEREF(__pyx_v_fp);
-          __pyx_t_3 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+          __pyx_t_3 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 69; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
           __Pyx_GOTREF(__pyx_t_3);
           __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
           __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
@@ -8102,7 +8063,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_18read_text(struct __pyx_obj_3_sa_Data
         __Pyx_XDECREF(__pyx_t_2); __pyx_t_2 = 0;
         __Pyx_XDECREF(__pyx_t_3); __pyx_t_3 = 0;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":71
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":68
  * 
  *     def read_text(self, char* filename):
  *         with gzip_or_text(filename) as fp:             # <<<<<<<<<<<<<<
@@ -8111,11 +8072,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_18read_text(struct __pyx_obj_3_sa_Data
  */
         /*except:*/ {
           __Pyx_AddTraceback("_sa.DataArray.read_text", __pyx_clineno, __pyx_lineno, __pyx_filename);
-          if (__Pyx_GetException(&__pyx_t_3, &__pyx_t_2, &__pyx_t_1) < 0) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (__Pyx_GetException(&__pyx_t_3, &__pyx_t_2, &__pyx_t_1) < 0) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __Pyx_GOTREF(__pyx_t_3);
           __Pyx_GOTREF(__pyx_t_2);
           __Pyx_GOTREF(__pyx_t_1);
-          __pyx_t_8 = PyTuple_New(3); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          __pyx_t_8 = PyTuple_New(3); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __Pyx_GOTREF(__pyx_t_8);
           __Pyx_INCREF(__pyx_t_3);
           PyTuple_SET_ITEM(__pyx_t_8, 0, __pyx_t_3);
@@ -8128,11 +8089,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_18read_text(struct __pyx_obj_3_sa_Data
           __Pyx_GIVEREF(__pyx_t_1);
           __pyx_t_10 = PyObject_Call(__pyx_t_4, __pyx_t_8, NULL);
           __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-          if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __Pyx_GOTREF(__pyx_t_10);
           __pyx_t_9 = __Pyx_PyObject_IsTrue(__pyx_t_10);
           __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
-          if (unlikely(__pyx_t_9 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (unlikely(__pyx_t_9 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __pyx_t_11 = (!__pyx_t_9);
           if (__pyx_t_11) {
             __Pyx_GIVEREF(__pyx_t_3);
@@ -8140,7 +8101,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_18read_text(struct __pyx_obj_3_sa_Data
             __Pyx_GIVEREF(__pyx_t_1);
             __Pyx_ErrRestore(__pyx_t_3, __pyx_t_2, __pyx_t_1);
             __pyx_t_3 = 0; __pyx_t_2 = 0; __pyx_t_1 = 0; 
-            {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+            {__pyx_filename = __pyx_f[3]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
             goto __pyx_L18;
           }
           __pyx_L18:;
@@ -8168,11 +8129,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_18read_text(struct __pyx_obj_3_sa_Data
       if (__pyx_t_4) {
         __pyx_t_7 = PyObject_Call(__pyx_t_4, __pyx_k_tuple_17, NULL);
         __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-        if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_7);
         __pyx_t_11 = __Pyx_PyObject_IsTrue(__pyx_t_7);
         __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
-        if (unlikely(__pyx_t_11 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        if (unlikely(__pyx_t_11 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       }
     }
     goto __pyx_L19;
@@ -8199,8 +8160,8 @@ static PyObject *__pyx_pf_3_sa_9DataArray_18read_text(struct __pyx_obj_3_sa_Data
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_9DataArray_21read_bitext(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static PyObject *__pyx_pw_3_sa_9DataArray_21read_bitext(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pw_3_sa_9DataArray_19read_bitext(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_19read_bitext(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   char *__pyx_v_filename;
   int __pyx_v_side;
   PyObject *__pyx_r = 0;
@@ -8226,11 +8187,11 @@ static PyObject *__pyx_pw_3_sa_9DataArray_21read_bitext(PyObject *__pyx_v_self,
         case  1:
         if (likely((values[1] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__side)) != 0)) kw_args--;
         else {
-          __Pyx_RaiseArgtupleInvalid("read_bitext", 1, 2, 2, 1); {__pyx_filename = __pyx_f[3]; __pyx_lineno = 74; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+          __Pyx_RaiseArgtupleInvalid("read_bitext", 1, 2, 2, 1); {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
         }
       }
       if (unlikely(kw_args > 0)) {
-        if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "read_bitext") < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 74; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+        if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "read_bitext") < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
     } else if (PyTuple_GET_SIZE(__pyx_args) != 2) {
       goto __pyx_L5_argtuple_error;
@@ -8238,24 +8199,24 @@ static PyObject *__pyx_pw_3_sa_9DataArray_21read_bitext(PyObject *__pyx_v_self,
       values[0] = PyTuple_GET_ITEM(__pyx_args, 0);
       values[1] = PyTuple_GET_ITEM(__pyx_args, 1);
     }
-    __pyx_v_filename = PyBytes_AsString(values[0]); if (unlikely((!__pyx_v_filename) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 74; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
-    __pyx_v_side = __Pyx_PyInt_AsInt(values[1]); if (unlikely((__pyx_v_side == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 74; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_v_filename = PyBytes_AsString(values[0]); if (unlikely((!__pyx_v_filename) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_v_side = __Pyx_PyInt_AsInt(values[1]); if (unlikely((__pyx_v_side == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("read_bitext", 1, 2, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[3]; __pyx_lineno = 74; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("read_bitext", 1, 2, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
   __Pyx_AddTraceback("_sa.DataArray.read_bitext", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_r = __pyx_pf_3_sa_9DataArray_20read_bitext(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), __pyx_v_filename, __pyx_v_side);
+  __pyx_r = __pyx_pf_3_sa_9DataArray_18read_bitext(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), __pyx_v_filename, __pyx_v_side);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 static PyObject *__pyx_gb_3_sa_9DataArray_11read_bitext_2generator6(__pyx_GeneratorObject *__pyx_generator, PyObject *__pyx_sent_value); /* proto */
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":76
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":73
  *     def read_bitext(self, char* filename, int side):
  *         with gzip_or_text(filename) as fp:
  *             data = (line.split(' ||| ')[side] for line in fp)             # <<<<<<<<<<<<<<
@@ -8281,7 +8242,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_11read_bitext_genexpr(PyObject *__pyx_
   __Pyx_INCREF(((PyObject *)__pyx_cur_scope->__pyx_outer_scope));
   __Pyx_GIVEREF(__pyx_cur_scope->__pyx_outer_scope);
   {
-    __pyx_GeneratorObject *gen = __Pyx_Generator_New((__pyx_generator_body_t) __pyx_gb_3_sa_9DataArray_11read_bitext_2generator6, (PyObject *) __pyx_cur_scope); if (unlikely(!gen)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 76; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_GeneratorObject *gen = __Pyx_Generator_New((__pyx_generator_body_t) __pyx_gb_3_sa_9DataArray_11read_bitext_2generator6, (PyObject *) __pyx_cur_scope); if (unlikely(!gen)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_cur_scope);
     __Pyx_RefNannyFinishContext();
     return (PyObject *) gen;
@@ -8318,13 +8279,13 @@ static PyObject *__pyx_gb_3_sa_9DataArray_11read_bitext_2generator6(__pyx_Genera
     return NULL;
   }
   __pyx_L3_first_run:;
-  if (unlikely(!__pyx_sent_value)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 76; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (unlikely(!__pyx_cur_scope->__pyx_outer_scope->__pyx_v_fp)) { __Pyx_RaiseClosureNameError("fp"); {__pyx_filename = __pyx_f[3]; __pyx_lineno = 76; __pyx_clineno = __LINE__; goto __pyx_L1_error;} }
+  if (unlikely(!__pyx_sent_value)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (unlikely(!__pyx_cur_scope->__pyx_outer_scope->__pyx_v_fp)) { __Pyx_RaiseClosureNameError("fp"); {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __pyx_clineno = __LINE__; goto __pyx_L1_error;} }
   if (PyList_CheckExact(__pyx_cur_scope->__pyx_outer_scope->__pyx_v_fp) || PyTuple_CheckExact(__pyx_cur_scope->__pyx_outer_scope->__pyx_v_fp)) {
     __pyx_t_1 = __pyx_cur_scope->__pyx_outer_scope->__pyx_v_fp; __Pyx_INCREF(__pyx_t_1); __pyx_t_2 = 0;
     __pyx_t_3 = NULL;
   } else {
-    __pyx_t_2 = -1; __pyx_t_1 = PyObject_GetIter(__pyx_cur_scope->__pyx_outer_scope->__pyx_v_fp); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 76; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = -1; __pyx_t_1 = PyObject_GetIter(__pyx_cur_scope->__pyx_outer_scope->__pyx_v_fp); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
     __pyx_t_3 = Py_TYPE(__pyx_t_1)->tp_iternext;
   }
@@ -8332,23 +8293,23 @@ static PyObject *__pyx_gb_3_sa_9DataArray_11read_bitext_2generator6(__pyx_Genera
     if (!__pyx_t_3 && PyList_CheckExact(__pyx_t_1)) {
       if (__pyx_t_2 >= PyList_GET_SIZE(__pyx_t_1)) break;
       #if CYTHON_COMPILING_IN_CPYTHON
-      __pyx_t_4 = PyList_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 76; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_4 = PyList_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #else
-      __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 76; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #endif
     } else if (!__pyx_t_3 && PyTuple_CheckExact(__pyx_t_1)) {
       if (__pyx_t_2 >= PyTuple_GET_SIZE(__pyx_t_1)) break;
       #if CYTHON_COMPILING_IN_CPYTHON
-      __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 76; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #else
-      __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 76; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #endif
     } else {
       __pyx_t_4 = __pyx_t_3(__pyx_t_1);
       if (unlikely(!__pyx_t_4)) {
         if (PyErr_Occurred()) {
           if (likely(PyErr_ExceptionMatches(PyExc_StopIteration))) PyErr_Clear();
-          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 76; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         }
         break;
       }
@@ -8359,12 +8320,12 @@ static PyObject *__pyx_gb_3_sa_9DataArray_11read_bitext_2generator6(__pyx_Genera
     __Pyx_GIVEREF(__pyx_t_4);
     __pyx_cur_scope->__pyx_v_line = __pyx_t_4;
     __pyx_t_4 = 0;
-    __pyx_t_4 = PyObject_GetAttr(__pyx_cur_scope->__pyx_v_line, __pyx_n_s__split); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 76; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyObject_GetAttr(__pyx_cur_scope->__pyx_v_line, __pyx_n_s__split); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
-    __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_k_tuple_19), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 76; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_k_tuple_19), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-    __pyx_t_4 = __Pyx_GetItemInt(__pyx_t_5, __pyx_cur_scope->__pyx_outer_scope->__pyx_v_side, sizeof(int), PyInt_FromLong); if (!__pyx_t_4) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 76; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = __Pyx_GetItemInt(__pyx_t_5, __pyx_cur_scope->__pyx_outer_scope->__pyx_v_side, sizeof(int), PyInt_FromLong); if (!__pyx_t_4) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
     __pyx_r = __pyx_t_4;
@@ -8384,7 +8345,7 @@ static PyObject *__pyx_gb_3_sa_9DataArray_11read_bitext_2generator6(__pyx_Genera
     __Pyx_XGOTREF(__pyx_t_1);
     __pyx_t_2 = __pyx_cur_scope->__pyx_t_1;
     __pyx_t_3 = __pyx_cur_scope->__pyx_t_2;
-    if (unlikely(!__pyx_sent_value)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 76; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    if (unlikely(!__pyx_sent_value)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   }
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
   PyErr_SetNone(PyExc_StopIteration);
@@ -8402,7 +8363,7 @@ static PyObject *__pyx_gb_3_sa_9DataArray_11read_bitext_2generator6(__pyx_Genera
   return NULL;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":74
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":71
  *             self.read_text_data(fp)
  * 
  *     def read_bitext(self, char* filename, int side):             # <<<<<<<<<<<<<<
@@ -8410,7 +8371,7 @@ static PyObject *__pyx_gb_3_sa_9DataArray_11read_bitext_2generator6(__pyx_Genera
  *             data = (line.split(' ||| ')[side] for line in fp)
  */
 
-static PyObject *__pyx_pf_3_sa_9DataArray_20read_bitext(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename, int __pyx_v_side) {
+static PyObject *__pyx_pf_3_sa_9DataArray_18read_bitext(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename, int __pyx_v_side) {
   struct __pyx_obj_3_sa___pyx_scope_struct_1_read_bitext *__pyx_cur_scope;
   PyObject *__pyx_v_data = NULL;
   PyObject *__pyx_r = NULL;
@@ -8438,7 +8399,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_bitext(struct __pyx_obj_3_sa_Da
   __Pyx_GOTREF(__pyx_cur_scope);
   __pyx_cur_scope->__pyx_v_side = __pyx_v_side;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":75
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":72
  * 
  *     def read_bitext(self, char* filename, int side):
  *         with gzip_or_text(filename) as fp:             # <<<<<<<<<<<<<<
@@ -8446,24 +8407,24 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_bitext(struct __pyx_obj_3_sa_Da
  *             self.read_text_data(data)
  */
   /*with:*/ {
-    __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s__gzip_or_text); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 75; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s__gzip_or_text); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
-    __pyx_t_2 = PyBytes_FromString(__pyx_v_filename); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 75; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyBytes_FromString(__pyx_v_filename); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_2));
-    __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 75; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
     PyTuple_SET_ITEM(__pyx_t_3, 0, ((PyObject *)__pyx_t_2));
     __Pyx_GIVEREF(((PyObject *)__pyx_t_2));
     __pyx_t_2 = 0;
-    __pyx_t_2 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 75; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
-    __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s____exit__); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 75; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s____exit__); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
-    __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s____enter__); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 75; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s____enter__); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     __Pyx_GOTREF(__pyx_t_3);
-    __pyx_t_1 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 75; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_t_1 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     __Pyx_GOTREF(__pyx_t_1);
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
@@ -8479,33 +8440,33 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_bitext(struct __pyx_obj_3_sa_Da
           __pyx_cur_scope->__pyx_v_fp = __pyx_t_1;
           __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":76
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":73
  *     def read_bitext(self, char* filename, int side):
  *         with gzip_or_text(filename) as fp:
  *             data = (line.split(' ||| ')[side] for line in fp)             # <<<<<<<<<<<<<<
  *             self.read_text_data(data)
  * 
  */
-          __pyx_t_1 = __pyx_pf_3_sa_9DataArray_11read_bitext_genexpr(((PyObject*)__pyx_cur_scope)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 76; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+          __pyx_t_1 = __pyx_pf_3_sa_9DataArray_11read_bitext_genexpr(((PyObject*)__pyx_cur_scope)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
           __Pyx_GOTREF(__pyx_t_1);
           __pyx_v_data = __pyx_t_1;
           __pyx_t_1 = 0;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":77
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":74
  *         with gzip_or_text(filename) as fp:
  *             data = (line.split(' ||| ')[side] for line in fp)
  *             self.read_text_data(data)             # <<<<<<<<<<<<<<
  * 
  *     def read_text_data(self, data):
  */
-          __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self), __pyx_n_s__read_text_data); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 77; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+          __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self), __pyx_n_s__read_text_data); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 74; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
           __Pyx_GOTREF(__pyx_t_1);
-          __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 77; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+          __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 74; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
           __Pyx_GOTREF(__pyx_t_2);
           __Pyx_INCREF(__pyx_v_data);
           PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_v_data);
           __Pyx_GIVEREF(__pyx_v_data);
-          __pyx_t_3 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 77; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+          __pyx_t_3 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 74; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
           __Pyx_GOTREF(__pyx_t_3);
           __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
           __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
@@ -8520,7 +8481,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_bitext(struct __pyx_obj_3_sa_Da
         __Pyx_XDECREF(__pyx_t_2); __pyx_t_2 = 0;
         __Pyx_XDECREF(__pyx_t_3); __pyx_t_3 = 0;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":75
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":72
  * 
  *     def read_bitext(self, char* filename, int side):
  *         with gzip_or_text(filename) as fp:             # <<<<<<<<<<<<<<
@@ -8529,11 +8490,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_bitext(struct __pyx_obj_3_sa_Da
  */
         /*except:*/ {
           __Pyx_AddTraceback("_sa.DataArray.read_bitext", __pyx_clineno, __pyx_lineno, __pyx_filename);
-          if (__Pyx_GetException(&__pyx_t_3, &__pyx_t_2, &__pyx_t_1) < 0) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 75; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (__Pyx_GetException(&__pyx_t_3, &__pyx_t_2, &__pyx_t_1) < 0) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __Pyx_GOTREF(__pyx_t_3);
           __Pyx_GOTREF(__pyx_t_2);
           __Pyx_GOTREF(__pyx_t_1);
-          __pyx_t_8 = PyTuple_New(3); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 75; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          __pyx_t_8 = PyTuple_New(3); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __Pyx_GOTREF(__pyx_t_8);
           __Pyx_INCREF(__pyx_t_3);
           PyTuple_SET_ITEM(__pyx_t_8, 0, __pyx_t_3);
@@ -8546,11 +8507,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_bitext(struct __pyx_obj_3_sa_Da
           __Pyx_GIVEREF(__pyx_t_1);
           __pyx_t_10 = PyObject_Call(__pyx_t_4, __pyx_t_8, NULL);
           __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-          if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 75; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __Pyx_GOTREF(__pyx_t_10);
           __pyx_t_9 = __Pyx_PyObject_IsTrue(__pyx_t_10);
           __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
-          if (unlikely(__pyx_t_9 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 75; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (unlikely(__pyx_t_9 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __pyx_t_11 = (!__pyx_t_9);
           if (__pyx_t_11) {
             __Pyx_GIVEREF(__pyx_t_3);
@@ -8558,7 +8519,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_bitext(struct __pyx_obj_3_sa_Da
             __Pyx_GIVEREF(__pyx_t_1);
             __Pyx_ErrRestore(__pyx_t_3, __pyx_t_2, __pyx_t_1);
             __pyx_t_3 = 0; __pyx_t_2 = 0; __pyx_t_1 = 0; 
-            {__pyx_filename = __pyx_f[3]; __pyx_lineno = 75; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+            {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
             goto __pyx_L18;
           }
           __pyx_L18:;
@@ -8586,11 +8547,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_bitext(struct __pyx_obj_3_sa_Da
       if (__pyx_t_4) {
         __pyx_t_7 = PyObject_Call(__pyx_t_4, __pyx_k_tuple_20, NULL);
         __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-        if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 75; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_7);
         __pyx_t_11 = __Pyx_PyObject_IsTrue(__pyx_t_7);
         __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
-        if (unlikely(__pyx_t_11 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 75; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        if (unlikely(__pyx_t_11 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       }
     }
     goto __pyx_L19;
@@ -8618,17 +8579,17 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_bitext(struct __pyx_obj_3_sa_Da
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_9DataArray_23read_text_data(PyObject *__pyx_v_self, PyObject *__pyx_v_data); /*proto*/
-static PyObject *__pyx_pw_3_sa_9DataArray_23read_text_data(PyObject *__pyx_v_self, PyObject *__pyx_v_data) {
+static PyObject *__pyx_pw_3_sa_9DataArray_21read_text_data(PyObject *__pyx_v_self, PyObject *__pyx_v_data); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_21read_text_data(PyObject *__pyx_v_self, PyObject *__pyx_v_data) {
   PyObject *__pyx_r = 0;
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("read_text_data (wrapper)", 0);
-  __pyx_r = __pyx_pf_3_sa_9DataArray_22read_text_data(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((PyObject *)__pyx_v_data));
+  __pyx_r = __pyx_pf_3_sa_9DataArray_20read_text_data(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((PyObject *)__pyx_v_data));
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":79
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":76
  *             self.read_text_data(data)
  * 
  *     def read_text_data(self, data):             # <<<<<<<<<<<<<<
@@ -8636,7 +8597,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_23read_text_data(PyObject *__pyx_v_sel
  *         for line_num, line in enumerate(data):
  */
 
-static PyObject *__pyx_pf_3_sa_9DataArray_22read_text_data(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_data) {
+static PyObject *__pyx_pf_3_sa_9DataArray_20read_text_data(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_data) {
   int __pyx_v_word_count;
   PyObject *__pyx_v_line_num = NULL;
   PyObject *__pyx_v_line = NULL;
@@ -8658,7 +8619,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_22read_text_data(struct __pyx_obj_3_sa
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("read_text_data", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":80
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":77
  * 
  *     def read_text_data(self, data):
  *         cdef int word_count = 0             # <<<<<<<<<<<<<<
@@ -8667,7 +8628,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_22read_text_data(struct __pyx_obj_3_sa
  */
   __pyx_v_word_count = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":81
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":78
  *     def read_text_data(self, data):
  *         cdef int word_count = 0
  *         for line_num, line in enumerate(data):             # <<<<<<<<<<<<<<
@@ -8680,7 +8641,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_22read_text_data(struct __pyx_obj_3_sa
     __pyx_t_2 = __pyx_v_data; __Pyx_INCREF(__pyx_t_2); __pyx_t_3 = 0;
     __pyx_t_4 = NULL;
   } else {
-    __pyx_t_3 = -1; __pyx_t_2 = PyObject_GetIter(__pyx_v_data); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 81; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = -1; __pyx_t_2 = PyObject_GetIter(__pyx_v_data); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 78; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_t_4 = Py_TYPE(__pyx_t_2)->tp_iternext;
   }
@@ -8688,23 +8649,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_22read_text_data(struct __pyx_obj_3_sa
     if (!__pyx_t_4 && PyList_CheckExact(__pyx_t_2)) {
       if (__pyx_t_3 >= PyList_GET_SIZE(__pyx_t_2)) break;
       #if CYTHON_COMPILING_IN_CPYTHON
-      __pyx_t_5 = PyList_GET_ITEM(__pyx_t_2, __pyx_t_3); __Pyx_INCREF(__pyx_t_5); __pyx_t_3++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 81; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_5 = PyList_GET_ITEM(__pyx_t_2, __pyx_t_3); __Pyx_INCREF(__pyx_t_5); __pyx_t_3++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 78; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #else
-      __pyx_t_5 = PySequence_ITEM(__pyx_t_2, __pyx_t_3); __pyx_t_3++; if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 81; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_5 = PySequence_ITEM(__pyx_t_2, __pyx_t_3); __pyx_t_3++; if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 78; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #endif
     } else if (!__pyx_t_4 && PyTuple_CheckExact(__pyx_t_2)) {
       if (__pyx_t_3 >= PyTuple_GET_SIZE(__pyx_t_2)) break;
       #if CYTHON_COMPILING_IN_CPYTHON
-      __pyx_t_5 = PyTuple_GET_ITEM(__pyx_t_2, __pyx_t_3); __Pyx_INCREF(__pyx_t_5); __pyx_t_3++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 81; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_5 = PyTuple_GET_ITEM(__pyx_t_2, __pyx_t_3); __Pyx_INCREF(__pyx_t_5); __pyx_t_3++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 78; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #else
-      __pyx_t_5 = PySequence_ITEM(__pyx_t_2, __pyx_t_3); __pyx_t_3++; if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 81; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_5 = PySequence_ITEM(__pyx_t_2, __pyx_t_3); __pyx_t_3++; if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 78; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #endif
     } else {
       __pyx_t_5 = __pyx_t_4(__pyx_t_2);
       if (unlikely(!__pyx_t_5)) {
         if (PyErr_Occurred()) {
           if (likely(PyErr_ExceptionMatches(PyExc_StopIteration))) PyErr_Clear();
-          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 81; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 78; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         }
         break;
       }
@@ -8716,43 +8677,43 @@ static PyObject *__pyx_pf_3_sa_9DataArray_22read_text_data(struct __pyx_obj_3_sa
     __Pyx_INCREF(__pyx_t_1);
     __Pyx_XDECREF(__pyx_v_line_num);
     __pyx_v_line_num = __pyx_t_1;
-    __pyx_t_5 = PyNumber_Add(__pyx_t_1, __pyx_int_1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 81; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyNumber_Add(__pyx_t_1, __pyx_int_1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 78; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
     __Pyx_DECREF(__pyx_t_1);
     __pyx_t_1 = __pyx_t_5;
     __pyx_t_5 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":82
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":79
  *         cdef int word_count = 0
  *         for line_num, line in enumerate(data):
  *             self.sent_index.append(word_count)             # <<<<<<<<<<<<<<
  *             for word in line.split():
  *                 self.data.append(self.get_id(word))
  */
-    __pyx_t_5 = PyInt_FromLong(__pyx_v_word_count); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 82; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyInt_FromLong(__pyx_v_word_count); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 79; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
-    __pyx_t_6 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->sent_index), __pyx_t_5); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 82; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_6 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->sent_index), __pyx_t_5); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 79; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_6);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
     __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":83
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":80
  *         for line_num, line in enumerate(data):
  *             self.sent_index.append(word_count)
  *             for word in line.split():             # <<<<<<<<<<<<<<
  *                 self.data.append(self.get_id(word))
  *                 if self.use_sent_id:
  */
-    __pyx_t_6 = PyObject_GetAttr(__pyx_v_line, __pyx_n_s__split); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 83; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_6 = PyObject_GetAttr(__pyx_v_line, __pyx_n_s__split); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 80; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_6);
-    __pyx_t_5 = PyObject_Call(__pyx_t_6, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 83; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyObject_Call(__pyx_t_6, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 80; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
     __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
     if (PyList_CheckExact(__pyx_t_5) || PyTuple_CheckExact(__pyx_t_5)) {
       __pyx_t_6 = __pyx_t_5; __Pyx_INCREF(__pyx_t_6); __pyx_t_7 = 0;
       __pyx_t_8 = NULL;
     } else {
-      __pyx_t_7 = -1; __pyx_t_6 = PyObject_GetIter(__pyx_t_5); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 83; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_7 = -1; __pyx_t_6 = PyObject_GetIter(__pyx_t_5); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 80; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_6);
       __pyx_t_8 = Py_TYPE(__pyx_t_6)->tp_iternext;
     }
@@ -8761,23 +8722,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_22read_text_data(struct __pyx_obj_3_sa
       if (!__pyx_t_8 && PyList_CheckExact(__pyx_t_6)) {
         if (__pyx_t_7 >= PyList_GET_SIZE(__pyx_t_6)) break;
         #if CYTHON_COMPILING_IN_CPYTHON
-        __pyx_t_5 = PyList_GET_ITEM(__pyx_t_6, __pyx_t_7); __Pyx_INCREF(__pyx_t_5); __pyx_t_7++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 83; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_5 = PyList_GET_ITEM(__pyx_t_6, __pyx_t_7); __Pyx_INCREF(__pyx_t_5); __pyx_t_7++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 80; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         #else
-        __pyx_t_5 = PySequence_ITEM(__pyx_t_6, __pyx_t_7); __pyx_t_7++; if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 83; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_5 = PySequence_ITEM(__pyx_t_6, __pyx_t_7); __pyx_t_7++; if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 80; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         #endif
       } else if (!__pyx_t_8 && PyTuple_CheckExact(__pyx_t_6)) {
         if (__pyx_t_7 >= PyTuple_GET_SIZE(__pyx_t_6)) break;
         #if CYTHON_COMPILING_IN_CPYTHON
-        __pyx_t_5 = PyTuple_GET_ITEM(__pyx_t_6, __pyx_t_7); __Pyx_INCREF(__pyx_t_5); __pyx_t_7++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 83; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_5 = PyTuple_GET_ITEM(__pyx_t_6, __pyx_t_7); __Pyx_INCREF(__pyx_t_5); __pyx_t_7++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 80; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         #else
-        __pyx_t_5 = PySequence_ITEM(__pyx_t_6, __pyx_t_7); __pyx_t_7++; if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 83; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_5 = PySequence_ITEM(__pyx_t_6, __pyx_t_7); __pyx_t_7++; if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 80; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         #endif
       } else {
         __pyx_t_5 = __pyx_t_8(__pyx_t_6);
         if (unlikely(!__pyx_t_5)) {
           if (PyErr_Occurred()) {
             if (likely(PyErr_ExceptionMatches(PyExc_StopIteration))) PyErr_Clear();
-            else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 83; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 80; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           }
           break;
         }
@@ -8787,30 +8748,30 @@ static PyObject *__pyx_pf_3_sa_9DataArray_22read_text_data(struct __pyx_obj_3_sa
       __pyx_v_word = __pyx_t_5;
       __pyx_t_5 = 0;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":84
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":81
  *             self.sent_index.append(word_count)
  *             for word in line.split():
  *                 self.data.append(self.get_id(word))             # <<<<<<<<<<<<<<
  *                 if self.use_sent_id:
  *                     self.sent_id.append(line_num)
  */
-      __pyx_t_5 = PyObject_GetAttr(((PyObject *)__pyx_v_self), __pyx_n_s__get_id); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 84; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_5 = PyObject_GetAttr(((PyObject *)__pyx_v_self), __pyx_n_s__get_id); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 81; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_5);
-      __pyx_t_9 = PyTuple_New(1); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 84; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_9 = PyTuple_New(1); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 81; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_9);
       __Pyx_INCREF(__pyx_v_word);
       PyTuple_SET_ITEM(__pyx_t_9, 0, __pyx_v_word);
       __Pyx_GIVEREF(__pyx_v_word);
-      __pyx_t_10 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_9), NULL); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 84; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_10 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_9), NULL); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 81; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_10);
       __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
       __Pyx_DECREF(((PyObject *)__pyx_t_9)); __pyx_t_9 = 0;
-      __pyx_t_9 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->data), __pyx_t_10); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 84; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_9 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->data), __pyx_t_10); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 81; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_9);
       __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
       __Pyx_DECREF(__pyx_t_9); __pyx_t_9 = 0;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":85
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":82
  *             for word in line.split():
  *                 self.data.append(self.get_id(word))
  *                 if self.use_sent_id:             # <<<<<<<<<<<<<<
@@ -8819,21 +8780,21 @@ static PyObject *__pyx_pf_3_sa_9DataArray_22read_text_data(struct __pyx_obj_3_sa
  */
       if (__pyx_v_self->use_sent_id) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":86
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":83
  *                 self.data.append(self.get_id(word))
  *                 if self.use_sent_id:
  *                     self.sent_id.append(line_num)             # <<<<<<<<<<<<<<
  *                 word_count = word_count + 1
  *             self.data.append(1)
  */
-        __pyx_t_9 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->sent_id), __pyx_v_line_num); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 86; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_9 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->sent_id), __pyx_v_line_num); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 83; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_9);
         __Pyx_DECREF(__pyx_t_9); __pyx_t_9 = 0;
         goto __pyx_L7;
       }
       __pyx_L7:;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":87
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":84
  *                 if self.use_sent_id:
  *                     self.sent_id.append(line_num)
  *                 word_count = word_count + 1             # <<<<<<<<<<<<<<
@@ -8844,18 +8805,18 @@ static PyObject *__pyx_pf_3_sa_9DataArray_22read_text_data(struct __pyx_obj_3_sa
     }
     __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":88
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":85
  *                     self.sent_id.append(line_num)
  *                 word_count = word_count + 1
  *             self.data.append(1)             # <<<<<<<<<<<<<<
  *             if self.use_sent_id:
  *                 self.sent_id.append(line_num)
  */
-    __pyx_t_6 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->data), __pyx_int_1); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 88; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_6 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->data), __pyx_int_1); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 85; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_6);
     __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":89
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":86
  *                 word_count = word_count + 1
  *             self.data.append(1)
  *             if self.use_sent_id:             # <<<<<<<<<<<<<<
@@ -8864,21 +8825,21 @@ static PyObject *__pyx_pf_3_sa_9DataArray_22read_text_data(struct __pyx_obj_3_sa
  */
     if (__pyx_v_self->use_sent_id) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":90
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":87
  *             self.data.append(1)
  *             if self.use_sent_id:
  *                 self.sent_id.append(line_num)             # <<<<<<<<<<<<<<
  *             word_count = word_count + 1
  *         self.data.append(0)
  */
-      __pyx_t_6 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->sent_id), __pyx_v_line_num); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 90; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_6 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->sent_id), __pyx_v_line_num); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 87; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_6);
       __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
       goto __pyx_L8;
     }
     __pyx_L8:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":91
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":88
  *             if self.use_sent_id:
  *                 self.sent_id.append(line_num)
  *             word_count = word_count + 1             # <<<<<<<<<<<<<<
@@ -8890,27 +8851,27 @@ static PyObject *__pyx_pf_3_sa_9DataArray_22read_text_data(struct __pyx_obj_3_sa
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":92
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":89
  *                 self.sent_id.append(line_num)
  *             word_count = word_count + 1
  *         self.data.append(0)             # <<<<<<<<<<<<<<
  *         self.sent_index.append(word_count)
  * 
  */
-  __pyx_t_1 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->data), __pyx_int_0); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 92; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->data), __pyx_int_0); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 89; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":93
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":90
  *             word_count = word_count + 1
  *         self.data.append(0)
  *         self.sent_index.append(word_count)             # <<<<<<<<<<<<<<
  * 
  * 
  */
-  __pyx_t_1 = PyInt_FromLong(__pyx_v_word_count); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 93; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyInt_FromLong(__pyx_v_word_count); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 90; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_2 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->sent_index), __pyx_t_1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 93; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->sent_index), __pyx_t_1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 90; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
@@ -8936,14 +8897,14 @@ static PyObject *__pyx_pf_3_sa_9DataArray_22read_text_data(struct __pyx_obj_3_sa
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_9DataArray_25read_binary(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename); /*proto*/
-static PyObject *__pyx_pw_3_sa_9DataArray_25read_binary(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename) {
+static PyObject *__pyx_pw_3_sa_9DataArray_23read_binary(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_23read_binary(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename) {
   char *__pyx_v_filename;
   PyObject *__pyx_r = 0;
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("read_binary (wrapper)", 0);
   assert(__pyx_arg_filename); {
-    __pyx_v_filename = PyBytes_AsString(__pyx_arg_filename); if (unlikely((!__pyx_v_filename) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 96; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_v_filename = PyBytes_AsString(__pyx_arg_filename); if (unlikely((!__pyx_v_filename) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 93; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L3_error:;
@@ -8951,12 +8912,12 @@ static PyObject *__pyx_pw_3_sa_9DataArray_25read_binary(PyObject *__pyx_v_self,
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_r = __pyx_pf_3_sa_9DataArray_24read_binary(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((char *)__pyx_v_filename));
+  __pyx_r = __pyx_pf_3_sa_9DataArray_22read_binary(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((char *)__pyx_v_filename));
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":96
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":93
  * 
  * 
  *     def read_binary(self, char* filename):             # <<<<<<<<<<<<<<
@@ -8964,13 +8925,13 @@ static PyObject *__pyx_pw_3_sa_9DataArray_25read_binary(PyObject *__pyx_v_self,
  *         f = fopen(filename, "r")
  */
 
-static PyObject *__pyx_pf_3_sa_9DataArray_24read_binary(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename) {
+static PyObject *__pyx_pf_3_sa_9DataArray_22read_binary(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename) {
   FILE *__pyx_v_f;
   PyObject *__pyx_r = NULL;
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("read_binary", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":98
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":95
  *     def read_binary(self, char* filename):
  *         cdef FILE* f
  *         f = fopen(filename, "r")             # <<<<<<<<<<<<<<
@@ -8979,7 +8940,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_24read_binary(struct __pyx_obj_3_sa_Da
  */
   __pyx_v_f = fopen(__pyx_v_filename, __pyx_k__r);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":99
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":96
  *         cdef FILE* f
  *         f = fopen(filename, "r")
  *         self.read_handle(f)             # <<<<<<<<<<<<<<
@@ -8988,7 +8949,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_24read_binary(struct __pyx_obj_3_sa_Da
  */
   ((struct __pyx_vtabstruct_3_sa_DataArray *)__pyx_v_self->__pyx_vtab)->read_handle(__pyx_v_self, __pyx_v_f);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":100
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":97
  *         f = fopen(filename, "r")
  *         self.read_handle(f)
  *         fclose(f)             # <<<<<<<<<<<<<<
@@ -9003,7 +8964,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_24read_binary(struct __pyx_obj_3_sa_Da
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":102
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":99
  *         fclose(f)
  * 
  *     cdef void read_handle(self, FILE* f):             # <<<<<<<<<<<<<<
@@ -9028,7 +8989,7 @@ static void __pyx_f_3_sa_9DataArray_read_handle(struct __pyx_obj_3_sa_DataArray
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("read_handle", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":107
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":104
  *         cdef char* word
  * 
  *         self.data.read_handle(f)             # <<<<<<<<<<<<<<
@@ -9037,7 +8998,7 @@ static void __pyx_f_3_sa_9DataArray_read_handle(struct __pyx_obj_3_sa_DataArray
  */
   ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_self->data->__pyx_vtab)->read_handle(__pyx_v_self->data, __pyx_v_f);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":108
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":105
  * 
  *         self.data.read_handle(f)
  *         self.sent_index.read_handle(f)             # <<<<<<<<<<<<<<
@@ -9046,7 +9007,7 @@ static void __pyx_f_3_sa_9DataArray_read_handle(struct __pyx_obj_3_sa_DataArray
  */
   ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_self->sent_index->__pyx_vtab)->read_handle(__pyx_v_self->sent_index, __pyx_v_f);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":109
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":106
  *         self.data.read_handle(f)
  *         self.sent_index.read_handle(f)
  *         self.sent_id.read_handle(f)             # <<<<<<<<<<<<<<
@@ -9055,7 +9016,7 @@ static void __pyx_f_3_sa_9DataArray_read_handle(struct __pyx_obj_3_sa_DataArray
  */
   ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_self->sent_id->__pyx_vtab)->read_handle(__pyx_v_self->sent_id, __pyx_v_f);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":110
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":107
  *         self.sent_index.read_handle(f)
  *         self.sent_id.read_handle(f)
  *         fread(&(num_words), sizeof(int), 1, f)             # <<<<<<<<<<<<<<
@@ -9064,7 +9025,7 @@ static void __pyx_f_3_sa_9DataArray_read_handle(struct __pyx_obj_3_sa_DataArray
  */
   fread((&__pyx_v_num_words), (sizeof(int)), 1, __pyx_v_f);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":111
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":108
  *         self.sent_id.read_handle(f)
  *         fread(&(num_words), sizeof(int), 1, f)
  *         for i in range(num_words):             # <<<<<<<<<<<<<<
@@ -9075,7 +9036,7 @@ static void __pyx_f_3_sa_9DataArray_read_handle(struct __pyx_obj_3_sa_DataArray
   for (__pyx_t_2 = 0; __pyx_t_2 < __pyx_t_1; __pyx_t_2+=1) {
     __pyx_v_i = __pyx_t_2;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":112
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":109
  *         fread(&(num_words), sizeof(int), 1, f)
  *         for i in range(num_words):
  *             fread(&(word_len), sizeof(int), 1, f)             # <<<<<<<<<<<<<<
@@ -9084,7 +9045,7 @@ static void __pyx_f_3_sa_9DataArray_read_handle(struct __pyx_obj_3_sa_DataArray
  */
     fread((&__pyx_v_word_len), (sizeof(int)), 1, __pyx_v_f);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":113
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":110
  *         for i in range(num_words):
  *             fread(&(word_len), sizeof(int), 1, f)
  *             word = <char*> malloc (word_len * sizeof(char))             # <<<<<<<<<<<<<<
@@ -9093,7 +9054,7 @@ static void __pyx_f_3_sa_9DataArray_read_handle(struct __pyx_obj_3_sa_DataArray
  */
     __pyx_v_word = ((char *)malloc((__pyx_v_word_len * (sizeof(char)))));
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":114
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":111
  *             fread(&(word_len), sizeof(int), 1, f)
  *             word = <char*> malloc (word_len * sizeof(char))
  *             fread(word, sizeof(char), word_len, f)             # <<<<<<<<<<<<<<
@@ -9102,7 +9063,7 @@ static void __pyx_f_3_sa_9DataArray_read_handle(struct __pyx_obj_3_sa_DataArray
  */
     fread(__pyx_v_word, (sizeof(char)), __pyx_v_word_len, __pyx_v_f);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":115
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":112
  *             word = <char*> malloc (word_len * sizeof(char))
  *             fread(word, sizeof(char), word_len, f)
  *             self.word2id[word] = len(self.id2word)             # <<<<<<<<<<<<<<
@@ -9111,31 +9072,31 @@ static void __pyx_f_3_sa_9DataArray_read_handle(struct __pyx_obj_3_sa_DataArray
  */
     __pyx_t_3 = __pyx_v_self->id2word;
     __Pyx_INCREF(__pyx_t_3);
-    __pyx_t_4 = PyObject_Length(__pyx_t_3); if (unlikely(__pyx_t_4 == -1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 115; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyObject_Length(__pyx_t_3); if (unlikely(__pyx_t_4 == -1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 112; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-    __pyx_t_3 = PyInt_FromSsize_t(__pyx_t_4); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 115; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyInt_FromSsize_t(__pyx_t_4); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 112; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
-    __pyx_t_5 = PyBytes_FromString(__pyx_v_word); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 115; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyBytes_FromString(__pyx_v_word); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 112; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_5));
-    if (PyObject_SetItem(__pyx_v_self->word2id, ((PyObject *)__pyx_t_5), __pyx_t_3) < 0) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 115; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    if (PyObject_SetItem(__pyx_v_self->word2id, ((PyObject *)__pyx_t_5), __pyx_t_3) < 0) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 112; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":116
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":113
  *             fread(word, sizeof(char), word_len, f)
  *             self.word2id[word] = len(self.id2word)
  *             self.id2word.append(word)             # <<<<<<<<<<<<<<
  *             free(word)
  *         if len(self.sent_id) == 0:
  */
-    __pyx_t_3 = PyBytes_FromString(__pyx_v_word); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 116; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyBytes_FromString(__pyx_v_word); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 113; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_3));
-    __pyx_t_5 = __Pyx_PyObject_Append(__pyx_v_self->id2word, ((PyObject *)__pyx_t_3)); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 116; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = __Pyx_PyObject_Append(__pyx_v_self->id2word, ((PyObject *)__pyx_t_3)); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 113; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
     __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":117
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":114
  *             self.word2id[word] = len(self.id2word)
  *             self.id2word.append(word)
  *             free(word)             # <<<<<<<<<<<<<<
@@ -9145,7 +9106,7 @@ static void __pyx_f_3_sa_9DataArray_read_handle(struct __pyx_obj_3_sa_DataArray
     free(__pyx_v_word);
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":118
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":115
  *             self.id2word.append(word)
  *             free(word)
  *         if len(self.sent_id) == 0:             # <<<<<<<<<<<<<<
@@ -9154,12 +9115,12 @@ static void __pyx_f_3_sa_9DataArray_read_handle(struct __pyx_obj_3_sa_DataArray
  */
   __pyx_t_5 = ((PyObject *)__pyx_v_self->sent_id);
   __Pyx_INCREF(__pyx_t_5);
-  __pyx_t_4 = PyObject_Length(__pyx_t_5); if (unlikely(__pyx_t_4 == -1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 118; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_Length(__pyx_t_5); if (unlikely(__pyx_t_4 == -1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 115; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   __pyx_t_6 = (__pyx_t_4 == 0);
   if (__pyx_t_6) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":119
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":116
  *             free(word)
  *         if len(self.sent_id) == 0:
  *             self.use_sent_id = False             # <<<<<<<<<<<<<<
@@ -9171,7 +9132,7 @@ static void __pyx_f_3_sa_9DataArray_read_handle(struct __pyx_obj_3_sa_DataArray
   }
   /*else*/ {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":121
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":118
  *             self.use_sent_id = False
  *         else:
  *             self.use_sent_id = True             # <<<<<<<<<<<<<<
@@ -9191,7 +9152,7 @@ static void __pyx_f_3_sa_9DataArray_read_handle(struct __pyx_obj_3_sa_DataArray
   __Pyx_RefNannyFinishContext();
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":123
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":120
  *             self.use_sent_id = True
  * 
  *     cdef void write_handle(self, FILE* f):             # <<<<<<<<<<<<<<
@@ -9215,7 +9176,7 @@ static void __pyx_f_3_sa_9DataArray_write_handle(struct __pyx_obj_3_sa_DataArray
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("write_handle", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":127
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":124
  *         cdef int num_words
  * 
  *         self.data.write_handle(f)             # <<<<<<<<<<<<<<
@@ -9224,7 +9185,7 @@ static void __pyx_f_3_sa_9DataArray_write_handle(struct __pyx_obj_3_sa_DataArray
  */
   ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_self->data->__pyx_vtab)->write_handle(__pyx_v_self->data, __pyx_v_f);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":128
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":125
  * 
  *         self.data.write_handle(f)
  *         self.sent_index.write_handle(f)             # <<<<<<<<<<<<<<
@@ -9233,7 +9194,7 @@ static void __pyx_f_3_sa_9DataArray_write_handle(struct __pyx_obj_3_sa_DataArray
  */
   ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_self->sent_index->__pyx_vtab)->write_handle(__pyx_v_self->sent_index, __pyx_v_f);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":129
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":126
  *         self.data.write_handle(f)
  *         self.sent_index.write_handle(f)
  *         self.sent_id.write_handle(f)             # <<<<<<<<<<<<<<
@@ -9242,7 +9203,7 @@ static void __pyx_f_3_sa_9DataArray_write_handle(struct __pyx_obj_3_sa_DataArray
  */
   ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_self->sent_id->__pyx_vtab)->write_handle(__pyx_v_self->sent_id, __pyx_v_f);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":130
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":127
  *         self.sent_index.write_handle(f)
  *         self.sent_id.write_handle(f)
  *         num_words = len(self.id2word) - 2             # <<<<<<<<<<<<<<
@@ -9251,11 +9212,11 @@ static void __pyx_f_3_sa_9DataArray_write_handle(struct __pyx_obj_3_sa_DataArray
  */
   __pyx_t_1 = __pyx_v_self->id2word;
   __Pyx_INCREF(__pyx_t_1);
-  __pyx_t_2 = PyObject_Length(__pyx_t_1); if (unlikely(__pyx_t_2 == -1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Length(__pyx_t_1); if (unlikely(__pyx_t_2 == -1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 127; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
   __pyx_v_num_words = (__pyx_t_2 - 2);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":131
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":128
  *         self.sent_id.write_handle(f)
  *         num_words = len(self.id2word) - 2
  *         fwrite(&(num_words), sizeof(int), 1, f)             # <<<<<<<<<<<<<<
@@ -9264,20 +9225,20 @@ static void __pyx_f_3_sa_9DataArray_write_handle(struct __pyx_obj_3_sa_DataArray
  */
   fwrite((&__pyx_v_num_words), (sizeof(int)), 1, __pyx_v_f);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":132
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":129
  *         num_words = len(self.id2word) - 2
  *         fwrite(&(num_words), sizeof(int), 1, f)
  *         for word in self.id2word[2:]:             # <<<<<<<<<<<<<<
  *             word_len = len(word) + 1
  *             fwrite(&(word_len), sizeof(int), 1, f)
  */
-  __pyx_t_1 = __Pyx_PySequence_GetSlice(__pyx_v_self->id2word, 2, PY_SSIZE_T_MAX); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PySequence_GetSlice(__pyx_v_self->id2word, 2, PY_SSIZE_T_MAX); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 129; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   if (PyList_CheckExact(__pyx_t_1) || PyTuple_CheckExact(__pyx_t_1)) {
     __pyx_t_3 = __pyx_t_1; __Pyx_INCREF(__pyx_t_3); __pyx_t_2 = 0;
     __pyx_t_4 = NULL;
   } else {
-    __pyx_t_2 = -1; __pyx_t_3 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = -1; __pyx_t_3 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 129; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
     __pyx_t_4 = Py_TYPE(__pyx_t_3)->tp_iternext;
   }
@@ -9286,23 +9247,23 @@ static void __pyx_f_3_sa_9DataArray_write_handle(struct __pyx_obj_3_sa_DataArray
     if (!__pyx_t_4 && PyList_CheckExact(__pyx_t_3)) {
       if (__pyx_t_2 >= PyList_GET_SIZE(__pyx_t_3)) break;
       #if CYTHON_COMPILING_IN_CPYTHON
-      __pyx_t_1 = PyList_GET_ITEM(__pyx_t_3, __pyx_t_2); __Pyx_INCREF(__pyx_t_1); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_1 = PyList_GET_ITEM(__pyx_t_3, __pyx_t_2); __Pyx_INCREF(__pyx_t_1); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 129; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #else
-      __pyx_t_1 = PySequence_ITEM(__pyx_t_3, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_1 = PySequence_ITEM(__pyx_t_3, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 129; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #endif
     } else if (!__pyx_t_4 && PyTuple_CheckExact(__pyx_t_3)) {
       if (__pyx_t_2 >= PyTuple_GET_SIZE(__pyx_t_3)) break;
       #if CYTHON_COMPILING_IN_CPYTHON
-      __pyx_t_1 = PyTuple_GET_ITEM(__pyx_t_3, __pyx_t_2); __Pyx_INCREF(__pyx_t_1); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_1 = PyTuple_GET_ITEM(__pyx_t_3, __pyx_t_2); __Pyx_INCREF(__pyx_t_1); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 129; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #else
-      __pyx_t_1 = PySequence_ITEM(__pyx_t_3, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_1 = PySequence_ITEM(__pyx_t_3, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 129; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #endif
     } else {
       __pyx_t_1 = __pyx_t_4(__pyx_t_3);
       if (unlikely(!__pyx_t_1)) {
         if (PyErr_Occurred()) {
           if (likely(PyErr_ExceptionMatches(PyExc_StopIteration))) PyErr_Clear();
-          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 129; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         }
         break;
       }
@@ -9312,17 +9273,17 @@ static void __pyx_f_3_sa_9DataArray_write_handle(struct __pyx_obj_3_sa_DataArray
     __pyx_v_word = __pyx_t_1;
     __pyx_t_1 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":133
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":130
  *         fwrite(&(num_words), sizeof(int), 1, f)
  *         for word in self.id2word[2:]:
  *             word_len = len(word) + 1             # <<<<<<<<<<<<<<
  *             fwrite(&(word_len), sizeof(int), 1, f)
  *             fwrite(<char *>word, sizeof(char), word_len, f)
  */
-    __pyx_t_5 = PyObject_Length(__pyx_v_word); if (unlikely(__pyx_t_5 == -1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 133; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyObject_Length(__pyx_v_word); if (unlikely(__pyx_t_5 == -1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __pyx_v_word_len = (__pyx_t_5 + 1);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":134
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":131
  *         for word in self.id2word[2:]:
  *             word_len = len(word) + 1
  *             fwrite(&(word_len), sizeof(int), 1, f)             # <<<<<<<<<<<<<<
@@ -9331,14 +9292,14 @@ static void __pyx_f_3_sa_9DataArray_write_handle(struct __pyx_obj_3_sa_DataArray
  */
     fwrite((&__pyx_v_word_len), (sizeof(int)), 1, __pyx_v_f);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":135
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":132
  *             word_len = len(word) + 1
  *             fwrite(&(word_len), sizeof(int), 1, f)
  *             fwrite(<char *>word, sizeof(char), word_len, f)             # <<<<<<<<<<<<<<
  * 
  *     def write_binary(self, char* filename):
  */
-    __pyx_t_6 = PyBytes_AsString(__pyx_v_word); if (unlikely((!__pyx_t_6) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 135; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_6 = PyBytes_AsString(__pyx_v_word); if (unlikely((!__pyx_t_6) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     fwrite(((char *)__pyx_t_6), (sizeof(char)), __pyx_v_word_len, __pyx_v_f);
   }
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
@@ -9354,14 +9315,14 @@ static void __pyx_f_3_sa_9DataArray_write_handle(struct __pyx_obj_3_sa_DataArray
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_9DataArray_27write_binary(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename); /*proto*/
-static PyObject *__pyx_pw_3_sa_9DataArray_27write_binary(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename) {
+static PyObject *__pyx_pw_3_sa_9DataArray_25write_binary(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_25write_binary(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename) {
   char *__pyx_v_filename;
   PyObject *__pyx_r = 0;
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("write_binary (wrapper)", 0);
   assert(__pyx_arg_filename); {
-    __pyx_v_filename = PyBytes_AsString(__pyx_arg_filename); if (unlikely((!__pyx_v_filename) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 137; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_v_filename = PyBytes_AsString(__pyx_arg_filename); if (unlikely((!__pyx_v_filename) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 134; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L3_error:;
@@ -9369,12 +9330,12 @@ static PyObject *__pyx_pw_3_sa_9DataArray_27write_binary(PyObject *__pyx_v_self,
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_r = __pyx_pf_3_sa_9DataArray_26write_binary(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((char *)__pyx_v_filename));
+  __pyx_r = __pyx_pf_3_sa_9DataArray_24write_binary(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((char *)__pyx_v_filename));
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":137
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":134
  *             fwrite(<char *>word, sizeof(char), word_len, f)
  * 
  *     def write_binary(self, char* filename):             # <<<<<<<<<<<<<<
@@ -9382,13 +9343,13 @@ static PyObject *__pyx_pw_3_sa_9DataArray_27write_binary(PyObject *__pyx_v_self,
  *         f = fopen(filename, "w")
  */
 
-static PyObject *__pyx_pf_3_sa_9DataArray_26write_binary(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename) {
+static PyObject *__pyx_pf_3_sa_9DataArray_24write_binary(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename) {
   FILE *__pyx_v_f;
   PyObject *__pyx_r = NULL;
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("write_binary", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":139
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":136
  *     def write_binary(self, char* filename):
  *         cdef FILE* f
  *         f = fopen(filename, "w")             # <<<<<<<<<<<<<<
@@ -9397,7 +9358,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_binary(struct __pyx_obj_3_sa_D
  */
   __pyx_v_f = fopen(__pyx_v_filename, __pyx_k__w);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":140
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":137
  *         cdef FILE* f
  *         f = fopen(filename, "w")
  *         self.write_handle(f)             # <<<<<<<<<<<<<<
@@ -9406,7 +9367,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_binary(struct __pyx_obj_3_sa_D
  */
   ((struct __pyx_vtabstruct_3_sa_DataArray *)__pyx_v_self->__pyx_vtab)->write_handle(__pyx_v_self, __pyx_v_f);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":141
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":138
  *         f = fopen(filename, "w")
  *         self.write_handle(f)
  *         fclose(f)             # <<<<<<<<<<<<<<
@@ -9422,17 +9383,17 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_binary(struct __pyx_obj_3_sa_D
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_9DataArray_29write_enhanced_handle(PyObject *__pyx_v_self, PyObject *__pyx_v_f); /*proto*/
-static PyObject *__pyx_pw_3_sa_9DataArray_29write_enhanced_handle(PyObject *__pyx_v_self, PyObject *__pyx_v_f) {
+static PyObject *__pyx_pw_3_sa_9DataArray_27write_enhanced_handle(PyObject *__pyx_v_self, PyObject *__pyx_v_f); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_27write_enhanced_handle(PyObject *__pyx_v_self, PyObject *__pyx_v_f) {
   PyObject *__pyx_r = 0;
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("write_enhanced_handle (wrapper)", 0);
-  __pyx_r = __pyx_pf_3_sa_9DataArray_28write_enhanced_handle(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((PyObject *)__pyx_v_f));
+  __pyx_r = __pyx_pf_3_sa_9DataArray_26write_enhanced_handle(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((PyObject *)__pyx_v_f));
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":143
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":140
  *         fclose(f)
  * 
  *     def write_enhanced_handle(self, f):             # <<<<<<<<<<<<<<
@@ -9440,7 +9401,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_29write_enhanced_handle(PyObject *__py
  *             f.write("%d " %i)
  */
 
-static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_f) {
+static PyObject *__pyx_pf_3_sa_9DataArray_26write_enhanced_handle(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_f) {
   PyObject *__pyx_v_i = NULL;
   PyObject *__pyx_v_word = NULL;
   PyObject *__pyx_r = NULL;
@@ -9456,7 +9417,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_o
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("write_enhanced_handle", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":144
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":141
  * 
  *     def write_enhanced_handle(self, f):
  *         for i in self.data:             # <<<<<<<<<<<<<<
@@ -9467,7 +9428,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_o
     __pyx_t_1 = ((PyObject *)__pyx_v_self->data); __Pyx_INCREF(__pyx_t_1); __pyx_t_2 = 0;
     __pyx_t_3 = NULL;
   } else {
-    __pyx_t_2 = -1; __pyx_t_1 = PyObject_GetIter(((PyObject *)__pyx_v_self->data)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 144; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = -1; __pyx_t_1 = PyObject_GetIter(((PyObject *)__pyx_v_self->data)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 141; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
     __pyx_t_3 = Py_TYPE(__pyx_t_1)->tp_iternext;
   }
@@ -9475,23 +9436,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_o
     if (!__pyx_t_3 && PyList_CheckExact(__pyx_t_1)) {
       if (__pyx_t_2 >= PyList_GET_SIZE(__pyx_t_1)) break;
       #if CYTHON_COMPILING_IN_CPYTHON
-      __pyx_t_4 = PyList_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 144; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_4 = PyList_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 141; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #else
-      __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 144; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 141; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #endif
     } else if (!__pyx_t_3 && PyTuple_CheckExact(__pyx_t_1)) {
       if (__pyx_t_2 >= PyTuple_GET_SIZE(__pyx_t_1)) break;
       #if CYTHON_COMPILING_IN_CPYTHON
-      __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 144; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 141; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #else
-      __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 144; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 141; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #endif
     } else {
       __pyx_t_4 = __pyx_t_3(__pyx_t_1);
       if (unlikely(!__pyx_t_4)) {
         if (PyErr_Occurred()) {
           if (likely(PyErr_ExceptionMatches(PyExc_StopIteration))) PyErr_Clear();
-          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 144; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 141; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         }
         break;
       }
@@ -9501,23 +9462,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_o
     __pyx_v_i = __pyx_t_4;
     __pyx_t_4 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":145
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":142
  *     def write_enhanced_handle(self, f):
  *         for i in self.data:
  *             f.write("%d " %i)             # <<<<<<<<<<<<<<
  *         f.write("\n")
  *         for i in self.sent_index:
  */
-    __pyx_t_4 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 142; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
-    __pyx_t_5 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_21), __pyx_v_i); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_21), __pyx_v_i); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 142; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_5));
-    __pyx_t_6 = PyTuple_New(1); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_6 = PyTuple_New(1); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 142; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_6);
     PyTuple_SET_ITEM(__pyx_t_6, 0, ((PyObject *)__pyx_t_5));
     __Pyx_GIVEREF(((PyObject *)__pyx_t_5));
     __pyx_t_5 = 0;
-    __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_6), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_6), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 142; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_6)); __pyx_t_6 = 0;
@@ -9525,21 +9486,21 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_o
   }
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":146
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":143
  *         for i in self.data:
  *             f.write("%d " %i)
  *         f.write("\n")             # <<<<<<<<<<<<<<
  *         for i in self.sent_index:
  *             f.write("%d " %i)
  */
-  __pyx_t_1 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 146; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 143; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_5 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_k_tuple_22), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 146; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_k_tuple_22), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 143; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":147
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":144
  *             f.write("%d " %i)
  *         f.write("\n")
  *         for i in self.sent_index:             # <<<<<<<<<<<<<<
@@ -9550,7 +9511,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_o
     __pyx_t_5 = ((PyObject *)__pyx_v_self->sent_index); __Pyx_INCREF(__pyx_t_5); __pyx_t_2 = 0;
     __pyx_t_3 = NULL;
   } else {
-    __pyx_t_2 = -1; __pyx_t_5 = PyObject_GetIter(((PyObject *)__pyx_v_self->sent_index)); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = -1; __pyx_t_5 = PyObject_GetIter(((PyObject *)__pyx_v_self->sent_index)); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 144; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
     __pyx_t_3 = Py_TYPE(__pyx_t_5)->tp_iternext;
   }
@@ -9558,23 +9519,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_o
     if (!__pyx_t_3 && PyList_CheckExact(__pyx_t_5)) {
       if (__pyx_t_2 >= PyList_GET_SIZE(__pyx_t_5)) break;
       #if CYTHON_COMPILING_IN_CPYTHON
-      __pyx_t_1 = PyList_GET_ITEM(__pyx_t_5, __pyx_t_2); __Pyx_INCREF(__pyx_t_1); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_1 = PyList_GET_ITEM(__pyx_t_5, __pyx_t_2); __Pyx_INCREF(__pyx_t_1); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 144; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #else
-      __pyx_t_1 = PySequence_ITEM(__pyx_t_5, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_1 = PySequence_ITEM(__pyx_t_5, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 144; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #endif
     } else if (!__pyx_t_3 && PyTuple_CheckExact(__pyx_t_5)) {
       if (__pyx_t_2 >= PyTuple_GET_SIZE(__pyx_t_5)) break;
       #if CYTHON_COMPILING_IN_CPYTHON
-      __pyx_t_1 = PyTuple_GET_ITEM(__pyx_t_5, __pyx_t_2); __Pyx_INCREF(__pyx_t_1); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_1 = PyTuple_GET_ITEM(__pyx_t_5, __pyx_t_2); __Pyx_INCREF(__pyx_t_1); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 144; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #else
-      __pyx_t_1 = PySequence_ITEM(__pyx_t_5, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_1 = PySequence_ITEM(__pyx_t_5, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 144; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #endif
     } else {
       __pyx_t_1 = __pyx_t_3(__pyx_t_5);
       if (unlikely(!__pyx_t_1)) {
         if (PyErr_Occurred()) {
           if (likely(PyErr_ExceptionMatches(PyExc_StopIteration))) PyErr_Clear();
-          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 144; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         }
         break;
       }
@@ -9584,23 +9545,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_o
     __pyx_v_i = __pyx_t_1;
     __pyx_t_1 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":148
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":145
  *         f.write("\n")
  *         for i in self.sent_index:
  *             f.write("%d " %i)             # <<<<<<<<<<<<<<
  *         f.write("\n")
  *         for i in self.sent_id:
  */
-    __pyx_t_1 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
-    __pyx_t_6 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_21), __pyx_v_i); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_6 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_21), __pyx_v_i); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_6));
-    __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
     PyTuple_SET_ITEM(__pyx_t_4, 0, ((PyObject *)__pyx_t_6));
     __Pyx_GIVEREF(((PyObject *)__pyx_t_6));
     __pyx_t_6 = 0;
-    __pyx_t_6 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_6 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_6);
     __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
@@ -9608,21 +9569,21 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_o
   }
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":149
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":146
  *         for i in self.sent_index:
  *             f.write("%d " %i)
  *         f.write("\n")             # <<<<<<<<<<<<<<
  *         for i in self.sent_id:
  *             f.write("%d " %i)
  */
-  __pyx_t_5 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 149; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 146; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
-  __pyx_t_6 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_k_tuple_23), NULL); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 149; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_6 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_k_tuple_23), NULL); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 146; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_6);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":150
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":147
  *             f.write("%d " %i)
  *         f.write("\n")
  *         for i in self.sent_id:             # <<<<<<<<<<<<<<
@@ -9633,7 +9594,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_o
     __pyx_t_6 = ((PyObject *)__pyx_v_self->sent_id); __Pyx_INCREF(__pyx_t_6); __pyx_t_2 = 0;
     __pyx_t_3 = NULL;
   } else {
-    __pyx_t_2 = -1; __pyx_t_6 = PyObject_GetIter(((PyObject *)__pyx_v_self->sent_id)); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 150; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = -1; __pyx_t_6 = PyObject_GetIter(((PyObject *)__pyx_v_self->sent_id)); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_6);
     __pyx_t_3 = Py_TYPE(__pyx_t_6)->tp_iternext;
   }
@@ -9641,23 +9602,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_o
     if (!__pyx_t_3 && PyList_CheckExact(__pyx_t_6)) {
       if (__pyx_t_2 >= PyList_GET_SIZE(__pyx_t_6)) break;
       #if CYTHON_COMPILING_IN_CPYTHON
-      __pyx_t_5 = PyList_GET_ITEM(__pyx_t_6, __pyx_t_2); __Pyx_INCREF(__pyx_t_5); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 150; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_5 = PyList_GET_ITEM(__pyx_t_6, __pyx_t_2); __Pyx_INCREF(__pyx_t_5); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #else
-      __pyx_t_5 = PySequence_ITEM(__pyx_t_6, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 150; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_5 = PySequence_ITEM(__pyx_t_6, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #endif
     } else if (!__pyx_t_3 && PyTuple_CheckExact(__pyx_t_6)) {
       if (__pyx_t_2 >= PyTuple_GET_SIZE(__pyx_t_6)) break;
       #if CYTHON_COMPILING_IN_CPYTHON
-      __pyx_t_5 = PyTuple_GET_ITEM(__pyx_t_6, __pyx_t_2); __Pyx_INCREF(__pyx_t_5); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 150; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_5 = PyTuple_GET_ITEM(__pyx_t_6, __pyx_t_2); __Pyx_INCREF(__pyx_t_5); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #else
-      __pyx_t_5 = PySequence_ITEM(__pyx_t_6, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 150; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_5 = PySequence_ITEM(__pyx_t_6, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #endif
     } else {
       __pyx_t_5 = __pyx_t_3(__pyx_t_6);
       if (unlikely(!__pyx_t_5)) {
         if (PyErr_Occurred()) {
           if (likely(PyErr_ExceptionMatches(PyExc_StopIteration))) PyErr_Clear();
-          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 150; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         }
         break;
       }
@@ -9667,23 +9628,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_o
     __pyx_v_i = __pyx_t_5;
     __pyx_t_5 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":151
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":148
  *         f.write("\n")
  *         for i in self.sent_id:
  *             f.write("%d " %i)             # <<<<<<<<<<<<<<
  *         f.write("\n")
  *         for word in self.id2word:
  */
-    __pyx_t_5 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 151; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
-    __pyx_t_4 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_21), __pyx_v_i); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 151; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_21), __pyx_v_i); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_4));
-    __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 151; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
     PyTuple_SET_ITEM(__pyx_t_1, 0, ((PyObject *)__pyx_t_4));
     __Pyx_GIVEREF(((PyObject *)__pyx_t_4));
     __pyx_t_4 = 0;
-    __pyx_t_4 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 151; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
@@ -9691,21 +9652,21 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_o
   }
   __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":152
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":149
  *         for i in self.sent_id:
  *             f.write("%d " %i)
  *         f.write("\n")             # <<<<<<<<<<<<<<
  *         for word in self.id2word:
  *             f.write("%s %d " % (word, self.word2id[word]))
  */
-  __pyx_t_6 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_6 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 149; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_6);
-  __pyx_t_4 = PyObject_Call(__pyx_t_6, ((PyObject *)__pyx_k_tuple_24), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_Call(__pyx_t_6, ((PyObject *)__pyx_k_tuple_24), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 149; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":153
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":150
  *             f.write("%d " %i)
  *         f.write("\n")
  *         for word in self.id2word:             # <<<<<<<<<<<<<<
@@ -9716,7 +9677,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_o
     __pyx_t_4 = __pyx_v_self->id2word; __Pyx_INCREF(__pyx_t_4); __pyx_t_2 = 0;
     __pyx_t_3 = NULL;
   } else {
-    __pyx_t_2 = -1; __pyx_t_4 = PyObject_GetIter(__pyx_v_self->id2word); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 153; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = -1; __pyx_t_4 = PyObject_GetIter(__pyx_v_self->id2word); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 150; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
     __pyx_t_3 = Py_TYPE(__pyx_t_4)->tp_iternext;
   }
@@ -9724,23 +9685,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_o
     if (!__pyx_t_3 && PyList_CheckExact(__pyx_t_4)) {
       if (__pyx_t_2 >= PyList_GET_SIZE(__pyx_t_4)) break;
       #if CYTHON_COMPILING_IN_CPYTHON
-      __pyx_t_6 = PyList_GET_ITEM(__pyx_t_4, __pyx_t_2); __Pyx_INCREF(__pyx_t_6); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 153; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_6 = PyList_GET_ITEM(__pyx_t_4, __pyx_t_2); __Pyx_INCREF(__pyx_t_6); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 150; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #else
-      __pyx_t_6 = PySequence_ITEM(__pyx_t_4, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 153; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_6 = PySequence_ITEM(__pyx_t_4, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 150; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #endif
     } else if (!__pyx_t_3 && PyTuple_CheckExact(__pyx_t_4)) {
       if (__pyx_t_2 >= PyTuple_GET_SIZE(__pyx_t_4)) break;
       #if CYTHON_COMPILING_IN_CPYTHON
-      __pyx_t_6 = PyTuple_GET_ITEM(__pyx_t_4, __pyx_t_2); __Pyx_INCREF(__pyx_t_6); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 153; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_6 = PyTuple_GET_ITEM(__pyx_t_4, __pyx_t_2); __Pyx_INCREF(__pyx_t_6); __pyx_t_2++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 150; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #else
-      __pyx_t_6 = PySequence_ITEM(__pyx_t_4, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 153; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_6 = PySequence_ITEM(__pyx_t_4, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 150; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       #endif
     } else {
       __pyx_t_6 = __pyx_t_3(__pyx_t_4);
       if (unlikely(!__pyx_t_6)) {
         if (PyErr_Occurred()) {
           if (likely(PyErr_ExceptionMatches(PyExc_StopIteration))) PyErr_Clear();
-          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 153; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 150; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         }
         break;
       }
@@ -9750,18 +9711,18 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_o
     __pyx_v_word = __pyx_t_6;
     __pyx_t_6 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":154
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":151
  *         f.write("\n")
  *         for word in self.id2word:
  *             f.write("%s %d " % (word, self.word2id[word]))             # <<<<<<<<<<<<<<
  *         f.write("\n")
  * 
  */
-    __pyx_t_6 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 154; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_6 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 151; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_6);
-    __pyx_t_1 = PyObject_GetItem(__pyx_v_self->word2id, __pyx_v_word); if (!__pyx_t_1) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 154; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = PyObject_GetItem(__pyx_v_self->word2id, __pyx_v_word); if (!__pyx_t_1) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 151; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
-    __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 154; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 151; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
     __Pyx_INCREF(__pyx_v_word);
     PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_v_word);
@@ -9769,15 +9730,15 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_o
     PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_1);
     __Pyx_GIVEREF(__pyx_t_1);
     __pyx_t_1 = 0;
-    __pyx_t_1 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_25), ((PyObject *)__pyx_t_5)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 154; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_25), ((PyObject *)__pyx_t_5)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 151; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_1));
     __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-    __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 154; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 151; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
     PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_t_1));
     __Pyx_GIVEREF(((PyObject *)__pyx_t_1));
     __pyx_t_1 = 0;
-    __pyx_t_1 = PyObject_Call(__pyx_t_6, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 154; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = PyObject_Call(__pyx_t_6, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 151; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
     __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
@@ -9785,16 +9746,16 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_o
   }
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":155
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":152
  *         for word in self.id2word:
  *             f.write("%s %d " % (word, self.word2id[word]))
  *         f.write("\n")             # <<<<<<<<<<<<<<
  * 
  *     def write_enhanced(self, char* filename):
  */
-  __pyx_t_4 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 155; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_k_tuple_26), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 155; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_k_tuple_26), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
@@ -9817,14 +9778,14 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced_handle(struct __pyx_o
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_9DataArray_31write_enhanced(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename); /*proto*/
-static PyObject *__pyx_pw_3_sa_9DataArray_31write_enhanced(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename) {
+static PyObject *__pyx_pw_3_sa_9DataArray_29write_enhanced(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_29write_enhanced(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename) {
   char *__pyx_v_filename;
   PyObject *__pyx_r = 0;
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("write_enhanced (wrapper)", 0);
   assert(__pyx_arg_filename); {
-    __pyx_v_filename = PyBytes_AsString(__pyx_arg_filename); if (unlikely((!__pyx_v_filename) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 157; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_v_filename = PyBytes_AsString(__pyx_arg_filename); if (unlikely((!__pyx_v_filename) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 154; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L3_error:;
@@ -9832,12 +9793,12 @@ static PyObject *__pyx_pw_3_sa_9DataArray_31write_enhanced(PyObject *__pyx_v_sel
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_r = __pyx_pf_3_sa_9DataArray_30write_enhanced(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((char *)__pyx_v_filename));
+  __pyx_r = __pyx_pf_3_sa_9DataArray_28write_enhanced(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((char *)__pyx_v_filename));
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":157
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":154
  *         f.write("\n")
  * 
  *     def write_enhanced(self, char* filename):             # <<<<<<<<<<<<<<
@@ -9845,7 +9806,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_31write_enhanced(PyObject *__pyx_v_sel
  *             self.write_enhanced_handle(self, f)
  */
 
-static PyObject *__pyx_pf_3_sa_9DataArray_30write_enhanced(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename) {
+static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, char *__pyx_v_filename) {
   PyObject *__pyx_v_f = NULL;
   PyObject *__pyx_r = NULL;
   __Pyx_RefNannyDeclarations
@@ -9865,16 +9826,16 @@ static PyObject *__pyx_pf_3_sa_9DataArray_30write_enhanced(struct __pyx_obj_3_sa
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("write_enhanced", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":158
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":155
  * 
  *     def write_enhanced(self, char* filename):
  *         with open(filename, "w") as f:             # <<<<<<<<<<<<<<
  *             self.write_enhanced_handle(self, f)
  */
   /*with:*/ {
-    __pyx_t_1 = PyBytes_FromString(__pyx_v_filename); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 158; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = PyBytes_FromString(__pyx_v_filename); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 155; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_1));
-    __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 158; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 155; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_t_1));
     __Pyx_GIVEREF(((PyObject *)__pyx_t_1));
@@ -9882,14 +9843,14 @@ static PyObject *__pyx_pf_3_sa_9DataArray_30write_enhanced(struct __pyx_obj_3_sa
     PyTuple_SET_ITEM(__pyx_t_2, 1, ((PyObject *)__pyx_n_s__w));
     __Pyx_GIVEREF(((PyObject *)__pyx_n_s__w));
     __pyx_t_1 = 0;
-    __pyx_t_1 = PyObject_Call(__pyx_builtin_open, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 158; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = PyObject_Call(__pyx_builtin_open, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 155; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
     __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
-    __pyx_t_3 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s____exit__); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 158; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s____exit__); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 155; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
-    __pyx_t_2 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s____enter__); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 158; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_t_2 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s____enter__); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 155; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_4 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 158; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_t_4 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 155; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     __Pyx_GOTREF(__pyx_t_4);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
     __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
@@ -9904,14 +9865,14 @@ static PyObject *__pyx_pf_3_sa_9DataArray_30write_enhanced(struct __pyx_obj_3_sa
           __pyx_v_f = __pyx_t_4;
           __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":159
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":156
  *     def write_enhanced(self, char* filename):
  *         with open(filename, "w") as f:
  *             self.write_enhanced_handle(self, f)             # <<<<<<<<<<<<<<
  */
-          __pyx_t_4 = PyObject_GetAttr(((PyObject *)__pyx_v_self), __pyx_n_s_27); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 159; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+          __pyx_t_4 = PyObject_GetAttr(((PyObject *)__pyx_v_self), __pyx_n_s_27); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 156; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
           __Pyx_GOTREF(__pyx_t_4);
-          __pyx_t_1 = PyTuple_New(2); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 159; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+          __pyx_t_1 = PyTuple_New(2); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 156; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
           __Pyx_GOTREF(__pyx_t_1);
           __Pyx_INCREF(((PyObject *)__pyx_v_self));
           PyTuple_SET_ITEM(__pyx_t_1, 0, ((PyObject *)__pyx_v_self));
@@ -9919,7 +9880,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_30write_enhanced(struct __pyx_obj_3_sa
           __Pyx_INCREF(__pyx_v_f);
           PyTuple_SET_ITEM(__pyx_t_1, 1, __pyx_v_f);
           __Pyx_GIVEREF(__pyx_v_f);
-          __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 159; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+          __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 156; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
           __Pyx_GOTREF(__pyx_t_2);
           __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
           __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
@@ -9934,7 +9895,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_30write_enhanced(struct __pyx_obj_3_sa
         __Pyx_XDECREF(__pyx_t_1); __pyx_t_1 = 0;
         __Pyx_XDECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":158
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":155
  * 
  *     def write_enhanced(self, char* filename):
  *         with open(filename, "w") as f:             # <<<<<<<<<<<<<<
@@ -9942,11 +9903,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_30write_enhanced(struct __pyx_obj_3_sa
  */
         /*except:*/ {
           __Pyx_AddTraceback("_sa.DataArray.write_enhanced", __pyx_clineno, __pyx_lineno, __pyx_filename);
-          if (__Pyx_GetException(&__pyx_t_2, &__pyx_t_1, &__pyx_t_4) < 0) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 158; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (__Pyx_GetException(&__pyx_t_2, &__pyx_t_1, &__pyx_t_4) < 0) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 155; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __Pyx_GOTREF(__pyx_t_2);
           __Pyx_GOTREF(__pyx_t_1);
           __Pyx_GOTREF(__pyx_t_4);
-          __pyx_t_8 = PyTuple_New(3); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 158; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          __pyx_t_8 = PyTuple_New(3); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 155; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __Pyx_GOTREF(__pyx_t_8);
           __Pyx_INCREF(__pyx_t_2);
           PyTuple_SET_ITEM(__pyx_t_8, 0, __pyx_t_2);
@@ -9959,11 +9920,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_30write_enhanced(struct __pyx_obj_3_sa
           __Pyx_GIVEREF(__pyx_t_4);
           __pyx_t_10 = PyObject_Call(__pyx_t_3, __pyx_t_8, NULL);
           __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-          if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 158; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 155; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __Pyx_GOTREF(__pyx_t_10);
           __pyx_t_9 = __Pyx_PyObject_IsTrue(__pyx_t_10);
           __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
-          if (unlikely(__pyx_t_9 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 158; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (unlikely(__pyx_t_9 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 155; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __pyx_t_11 = (!__pyx_t_9);
           if (__pyx_t_11) {
             __Pyx_GIVEREF(__pyx_t_2);
@@ -9971,7 +9932,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_30write_enhanced(struct __pyx_obj_3_sa
             __Pyx_GIVEREF(__pyx_t_4);
             __Pyx_ErrRestore(__pyx_t_2, __pyx_t_1, __pyx_t_4);
             __pyx_t_2 = 0; __pyx_t_1 = 0; __pyx_t_4 = 0; 
-            {__pyx_filename = __pyx_f[3]; __pyx_lineno = 158; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+            {__pyx_filename = __pyx_f[3]; __pyx_lineno = 155; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
             goto __pyx_L18;
           }
           __pyx_L18:;
@@ -9999,11 +9960,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_30write_enhanced(struct __pyx_obj_3_sa
       if (__pyx_t_3) {
         __pyx_t_7 = PyObject_Call(__pyx_t_3, __pyx_k_tuple_28, NULL);
         __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-        if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 158; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 155; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_7);
         __pyx_t_11 = __Pyx_PyObject_IsTrue(__pyx_t_7);
         __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
-        if (unlikely(__pyx_t_11 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 158; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        if (unlikely(__pyx_t_11 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 155; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       }
     }
     goto __pyx_L19;
@@ -10029,14 +9990,102 @@ static PyObject *__pyx_pf_3_sa_9DataArray_30write_enhanced(struct __pyx_obj_3_sa
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/alignment.pxi":12
+/* Python wrapper */
+static PyObject *__pyx_pw_3_sa_9DataArray_4data_1__get__(PyObject *__pyx_v_self); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_4data_1__get__(PyObject *__pyx_v_self) {
+  PyObject *__pyx_r = 0;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__get__ (wrapper)", 0);
+  __pyx_r = __pyx_pf_3_sa_9DataArray_4data___get__(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self));
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":12
+ *     cdef word2id
+ *     cdef id2word
+ *     cdef public IntList data             # <<<<<<<<<<<<<<
+ *     cdef IntList sent_id
  *     cdef IntList sent_index
- * 
- *     cdef int link(self, int i, int j):             # <<<<<<<<<<<<<<
- *         """Integerizes an alignment link pair"""
- *         return i*65536 + j
  */
 
+static PyObject *__pyx_pf_3_sa_9DataArray_4data___get__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self) {
+  PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__get__", 0);
+  __Pyx_XDECREF(__pyx_r);
+  __Pyx_INCREF(((PyObject *)__pyx_v_self->data));
+  __pyx_r = ((PyObject *)__pyx_v_self->data);
+  goto __pyx_L0;
+
+  __pyx_r = Py_None; __Pyx_INCREF(Py_None);
+  __pyx_L0:;
+  __Pyx_XGIVEREF(__pyx_r);
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+/* Python wrapper */
+static int __pyx_pw_3_sa_9DataArray_4data_3__set__(PyObject *__pyx_v_self, PyObject *__pyx_v_value); /*proto*/
+static int __pyx_pw_3_sa_9DataArray_4data_3__set__(PyObject *__pyx_v_self, PyObject *__pyx_v_value) {
+  int __pyx_r;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__set__ (wrapper)", 0);
+  __pyx_r = __pyx_pf_3_sa_9DataArray_4data_2__set__(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((PyObject *)__pyx_v_value));
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+static int __pyx_pf_3_sa_9DataArray_4data_2__set__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_value) {
+  int __pyx_r;
+  __Pyx_RefNannyDeclarations
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
+  __Pyx_RefNannySetupContext("__set__", 0);
+  if (!(likely(((__pyx_v_value) == Py_None) || likely(__Pyx_TypeTest(__pyx_v_value, __pyx_ptype_3_sa_IntList))))) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 12; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_INCREF(__pyx_v_value);
+  __Pyx_GIVEREF(__pyx_v_value);
+  __Pyx_GOTREF(__pyx_v_self->data);
+  __Pyx_DECREF(((PyObject *)__pyx_v_self->data));
+  __pyx_v_self->data = ((struct __pyx_obj_3_sa_IntList *)__pyx_v_value);
+
+  __pyx_r = 0;
+  goto __pyx_L0;
+  __pyx_L1_error:;
+  __Pyx_AddTraceback("_sa.DataArray.data.__set__", __pyx_clineno, __pyx_lineno, __pyx_filename);
+  __pyx_r = -1;
+  __pyx_L0:;
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+/* Python wrapper */
+static int __pyx_pw_3_sa_9DataArray_4data_5__del__(PyObject *__pyx_v_self); /*proto*/
+static int __pyx_pw_3_sa_9DataArray_4data_5__del__(PyObject *__pyx_v_self) {
+  int __pyx_r;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__del__ (wrapper)", 0);
+  __pyx_r = __pyx_pf_3_sa_9DataArray_4data_4__del__(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self));
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+static int __pyx_pf_3_sa_9DataArray_4data_4__del__(struct __pyx_obj_3_sa_DataArray *__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->data);
+  __Pyx_DECREF(((PyObject *)__pyx_v_self->data));
+  __pyx_v_self->data = ((struct __pyx_obj_3_sa_IntList *)Py_None);
+
+  __pyx_r = 0;
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
 static int __pyx_f_3_sa_9Alignment_link(CYTHON_UNUSED struct __pyx_obj_3_sa_Alignment *__pyx_v_self, int __pyx_v_i, int __pyx_v_j) {
   int __pyx_r;
   __Pyx_RefNannyDeclarations
@@ -45617,8 +45666,8 @@ static PyObject *__pyx_pw_3_sa_23HieroCachingRuleFactory_5input_1lambda3(PyObjec
  *                             for f, elist in fphrases.iteritems():
  *                                 for e, alslist in elist.iteritems():
  *                                     alignment, max_locs = max(alslist.iteritems(), key=lambda x: len(x[1]))             # <<<<<<<<<<<<<<
- *                                     locs = tuple(itertools.chain(alslist.itervalues()))
- *                                     # count = len(locs) # Should be?
+ *                                     locs = tuple(itertools.chain.from_iterable(alslist.itervalues()))
+ *                                     count = len(locs)
  */
 
 static PyObject *__pyx_lambda_funcdef_lambda3(CYTHON_UNUSED PyObject *__pyx_self, PyObject *__pyx_v_x) {
@@ -47931,7 +47980,7 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
  *                             for f, elist in fphrases.iteritems():
  *                                 for e, alslist in elist.iteritems():             # <<<<<<<<<<<<<<
  *                                     alignment, max_locs = max(alslist.iteritems(), key=lambda x: len(x[1]))
- *                                     locs = tuple(itertools.chain(alslist.itervalues()))
+ *                                     locs = tuple(itertools.chain.from_iterable(alslist.itervalues()))
  */
               __pyx_t_23 = 0;
               if (unlikely(__pyx_cur_scope->__pyx_v_elist == Py_None)) {
@@ -47964,8 +48013,8 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
  *                             for f, elist in fphrases.iteritems():
  *                                 for e, alslist in elist.iteritems():
  *                                     alignment, max_locs = max(alslist.iteritems(), key=lambda x: len(x[1]))             # <<<<<<<<<<<<<<
- *                                     locs = tuple(itertools.chain(alslist.itervalues()))
- *                                     # count = len(locs) # Should be?
+ *                                     locs = tuple(itertools.chain.from_iterable(alslist.itervalues()))
+ *                                     count = len(locs)
  */
                 __pyx_t_14 = PyObject_GetAttr(__pyx_cur_scope->__pyx_v_alslist, __pyx_n_s__iteritems); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1103; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_14);
@@ -48050,52 +48099,55 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
                 /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1104
  *                                 for e, alslist in elist.iteritems():
  *                                     alignment, max_locs = max(alslist.iteritems(), key=lambda x: len(x[1]))
- *                                     locs = tuple(itertools.chain(alslist.itervalues()))             # <<<<<<<<<<<<<<
- *                                     # count = len(locs) # Should be?
- *                                     count = len(max_locs) # Was
+ *                                     locs = tuple(itertools.chain.from_iterable(alslist.itervalues()))             # <<<<<<<<<<<<<<
+ *                                     count = len(locs)
+ *                                     scores = self.scorer.score(FeatureContext(
  */
                 __pyx_t_13 = __Pyx_GetName(__pyx_m, __pyx_n_s__itertools); if (unlikely(!__pyx_t_13)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1104; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_13);
                 __pyx_t_14 = PyObject_GetAttr(__pyx_t_13, __pyx_n_s__chain); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1104; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_14);
                 __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0;
-                __pyx_t_13 = PyObject_GetAttr(__pyx_cur_scope->__pyx_v_alslist, __pyx_n_s__itervalues); if (unlikely(!__pyx_t_13)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1104; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_13 = PyObject_GetAttr(__pyx_t_14, __pyx_n_s__from_iterable); if (unlikely(!__pyx_t_13)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1104; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_13);
-                __pyx_t_9 = PyObject_Call(__pyx_t_13, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1104; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
+                __pyx_t_14 = PyObject_GetAttr(__pyx_cur_scope->__pyx_v_alslist, __pyx_n_s__itervalues); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1104; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __Pyx_GOTREF(__pyx_t_14);
+                __pyx_t_9 = PyObject_Call(__pyx_t_14, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1104; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_9);
-                __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0;
-                __pyx_t_13 = PyTuple_New(1); if (unlikely(!__pyx_t_13)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1104; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-                __Pyx_GOTREF(__pyx_t_13);
-                PyTuple_SET_ITEM(__pyx_t_13, 0, __pyx_t_9);
+                __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
+                __pyx_t_14 = PyTuple_New(1); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1104; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __Pyx_GOTREF(__pyx_t_14);
+                PyTuple_SET_ITEM(__pyx_t_14, 0, __pyx_t_9);
                 __Pyx_GIVEREF(__pyx_t_9);
                 __pyx_t_9 = 0;
-                __pyx_t_9 = PyObject_Call(__pyx_t_14, ((PyObject *)__pyx_t_13), NULL); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1104; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_9 = PyObject_Call(__pyx_t_13, ((PyObject *)__pyx_t_14), NULL); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1104; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_9);
-                __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
-                __Pyx_DECREF(((PyObject *)__pyx_t_13)); __pyx_t_13 = 0;
-                __pyx_t_13 = PyTuple_New(1); if (unlikely(!__pyx_t_13)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1104; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-                __Pyx_GOTREF(__pyx_t_13);
-                PyTuple_SET_ITEM(__pyx_t_13, 0, __pyx_t_9);
+                __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0;
+                __Pyx_DECREF(((PyObject *)__pyx_t_14)); __pyx_t_14 = 0;
+                __pyx_t_14 = PyTuple_New(1); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1104; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __Pyx_GOTREF(__pyx_t_14);
+                PyTuple_SET_ITEM(__pyx_t_14, 0, __pyx_t_9);
                 __Pyx_GIVEREF(__pyx_t_9);
                 __pyx_t_9 = 0;
-                __pyx_t_9 = PyObject_Call(((PyObject *)((PyObject*)(&PyTuple_Type))), ((PyObject *)__pyx_t_13), NULL); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1104; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_9 = PyObject_Call(((PyObject *)((PyObject*)(&PyTuple_Type))), ((PyObject *)__pyx_t_14), NULL); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1104; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_9);
-                __Pyx_DECREF(((PyObject *)__pyx_t_13)); __pyx_t_13 = 0;
+                __Pyx_DECREF(((PyObject *)__pyx_t_14)); __pyx_t_14 = 0;
                 __Pyx_XGOTREF(((PyObject *)__pyx_cur_scope->__pyx_v_locs));
                 __Pyx_XDECREF(((PyObject *)__pyx_cur_scope->__pyx_v_locs));
                 __Pyx_GIVEREF(__pyx_t_9);
                 __pyx_cur_scope->__pyx_v_locs = ((PyObject*)__pyx_t_9);
                 __pyx_t_9 = 0;
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1106
- *                                     locs = tuple(itertools.chain(alslist.itervalues()))
- *                                     # count = len(locs) # Should be?
- *                                     count = len(max_locs) # Was             # <<<<<<<<<<<<<<
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1105
+ *                                     alignment, max_locs = max(alslist.iteritems(), key=lambda x: len(x[1]))
+ *                                     locs = tuple(itertools.chain.from_iterable(alslist.itervalues()))
+ *                                     count = len(locs)             # <<<<<<<<<<<<<<
  *                                     scores = self.scorer.score(FeatureContext(
  *                                                f, e, count, fcount[f], num_samples,
  */
-                __pyx_t_25 = PyObject_Length(__pyx_cur_scope->__pyx_v_max_locs); if (unlikely(__pyx_t_25 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1106; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-                __pyx_t_9 = PyInt_FromSsize_t(__pyx_t_25); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1106; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_25 = PyTuple_GET_SIZE(((PyObject *)__pyx_cur_scope->__pyx_v_locs)); if (unlikely(__pyx_t_25 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1105; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_9 = PyInt_FromSsize_t(__pyx_t_25); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1105; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_9);
                 __Pyx_XGOTREF(__pyx_cur_scope->__pyx_v_count);
                 __Pyx_XDECREF(__pyx_cur_scope->__pyx_v_count);
@@ -48103,43 +48155,43 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
                 __pyx_cur_scope->__pyx_v_count = __pyx_t_9;
                 __pyx_t_9 = 0;
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1107
- *                                     # count = len(locs) # Should be?
- *                                     count = len(max_locs) # Was
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1106
+ *                                     locs = tuple(itertools.chain.from_iterable(alslist.itervalues()))
+ *                                     count = len(locs)
  *                                     scores = self.scorer.score(FeatureContext(             # <<<<<<<<<<<<<<
  *                                                f, e, count, fcount[f], num_samples,
  *                                                (k,i+spanlen), locs, fwords, self.fda, self.eda
  */
-                __pyx_t_9 = __Pyx_GetName(__pyx_m, __pyx_n_s__FeatureContext); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_9 = __Pyx_GetName(__pyx_m, __pyx_n_s__FeatureContext); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1106; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_9);
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1108
- *                                     count = len(max_locs) # Was
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1107
+ *                                     count = len(locs)
  *                                     scores = self.scorer.score(FeatureContext(
  *                                                f, e, count, fcount[f], num_samples,             # <<<<<<<<<<<<<<
  *                                                (k,i+spanlen), locs, fwords, self.fda, self.eda
  *                                                ))
  */
-                __pyx_t_13 = PyObject_GetItem(__pyx_cur_scope->__pyx_v_fcount, __pyx_cur_scope->__pyx_v_f); if (!__pyx_t_13) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1108; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-                __Pyx_GOTREF(__pyx_t_13);
-                __pyx_t_14 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_num_samples); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1108; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_14 = PyObject_GetItem(__pyx_cur_scope->__pyx_v_fcount, __pyx_cur_scope->__pyx_v_f); if (!__pyx_t_14) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_14);
+                __pyx_t_13 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_num_samples); if (unlikely(!__pyx_t_13)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __Pyx_GOTREF(__pyx_t_13);
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1109
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1108
  *                                     scores = self.scorer.score(FeatureContext(
  *                                                f, e, count, fcount[f], num_samples,
  *                                                (k,i+spanlen), locs, fwords, self.fda, self.eda             # <<<<<<<<<<<<<<
  *                                                ))
  *                                     yield Rule(self.category, f, e, scores, alignment)
  */
-                __pyx_t_7 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_k); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1109; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_7 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_k); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1108; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_7);
-                __pyx_t_2 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_i); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1109; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_2 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_i); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1108; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_2);
-                __pyx_t_15 = PyNumber_Add(__pyx_t_2, __pyx_cur_scope->__pyx_v_spanlen); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1109; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_15 = PyNumber_Add(__pyx_t_2, __pyx_cur_scope->__pyx_v_spanlen); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1108; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_15);
                 __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-                __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1109; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1108; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_2);
                 PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_t_7);
                 __Pyx_GIVEREF(__pyx_t_7);
@@ -48147,7 +48199,7 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
                 __Pyx_GIVEREF(__pyx_t_15);
                 __pyx_t_7 = 0;
                 __pyx_t_15 = 0;
-                __pyx_t_15 = PyTuple_New(10); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_15 = PyTuple_New(10); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1106; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_15);
                 __Pyx_INCREF(__pyx_cur_scope->__pyx_v_f);
                 PyTuple_SET_ITEM(__pyx_t_15, 0, __pyx_cur_scope->__pyx_v_f);
@@ -48158,10 +48210,10 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
                 __Pyx_INCREF(__pyx_cur_scope->__pyx_v_count);
                 PyTuple_SET_ITEM(__pyx_t_15, 2, __pyx_cur_scope->__pyx_v_count);
                 __Pyx_GIVEREF(__pyx_cur_scope->__pyx_v_count);
-                PyTuple_SET_ITEM(__pyx_t_15, 3, __pyx_t_13);
-                __Pyx_GIVEREF(__pyx_t_13);
-                PyTuple_SET_ITEM(__pyx_t_15, 4, __pyx_t_14);
+                PyTuple_SET_ITEM(__pyx_t_15, 3, __pyx_t_14);
                 __Pyx_GIVEREF(__pyx_t_14);
+                PyTuple_SET_ITEM(__pyx_t_15, 4, __pyx_t_13);
+                __Pyx_GIVEREF(__pyx_t_13);
                 PyTuple_SET_ITEM(__pyx_t_15, 5, ((PyObject *)__pyx_t_2));
                 __Pyx_GIVEREF(((PyObject *)__pyx_t_2));
                 __Pyx_INCREF(((PyObject *)__pyx_cur_scope->__pyx_v_locs));
@@ -48176,14 +48228,14 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
                 __Pyx_INCREF(((PyObject *)__pyx_cur_scope->__pyx_v_self->eda));
                 PyTuple_SET_ITEM(__pyx_t_15, 9, ((PyObject *)__pyx_cur_scope->__pyx_v_self->eda));
                 __Pyx_GIVEREF(((PyObject *)__pyx_cur_scope->__pyx_v_self->eda));
-                __pyx_t_13 = 0;
                 __pyx_t_14 = 0;
+                __pyx_t_13 = 0;
                 __pyx_t_2 = 0;
-                __pyx_t_2 = PyObject_Call(__pyx_t_9, ((PyObject *)__pyx_t_15), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_2 = PyObject_Call(__pyx_t_9, ((PyObject *)__pyx_t_15), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1106; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_2);
                 __Pyx_DECREF(__pyx_t_9); __pyx_t_9 = 0;
                 __Pyx_DECREF(((PyObject *)__pyx_t_15)); __pyx_t_15 = 0;
-                __pyx_t_15 = ((PyObject *)((struct __pyx_vtabstruct_3_sa_Scorer *)__pyx_cur_scope->__pyx_v_self->scorer->__pyx_vtab)->score(__pyx_cur_scope->__pyx_v_self->scorer, __pyx_t_2)); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_15 = ((PyObject *)((struct __pyx_vtabstruct_3_sa_Scorer *)__pyx_cur_scope->__pyx_v_self->scorer->__pyx_vtab)->score(__pyx_cur_scope->__pyx_v_self->scorer, __pyx_t_2)); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1106; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_15);
                 __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
                 __Pyx_XGOTREF(((PyObject *)__pyx_cur_scope->__pyx_v_scores));
@@ -48192,16 +48244,16 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
                 __pyx_cur_scope->__pyx_v_scores = ((struct __pyx_obj_3_sa_FeatureVector *)__pyx_t_15);
                 __pyx_t_15 = 0;
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1111
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1110
  *                                                (k,i+spanlen), locs, fwords, self.fda, self.eda
  *                                                ))
  *                                     yield Rule(self.category, f, e, scores, alignment)             # <<<<<<<<<<<<<<
  * 
  *                 if len(phrase) < self.max_length and i+spanlen < len(fwords) and pathlen+1 <= self.max_initial_size:
  */
-                __pyx_t_15 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_self->category); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1111; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_15 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_self->category); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1110; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_15);
-                __pyx_t_2 = PyTuple_New(5); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1111; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_2 = PyTuple_New(5); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1110; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_2);
                 PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_t_15);
                 __Pyx_GIVEREF(__pyx_t_15);
@@ -48218,7 +48270,7 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
                 PyTuple_SET_ITEM(__pyx_t_2, 4, __pyx_cur_scope->__pyx_v_alignment);
                 __Pyx_GIVEREF(__pyx_cur_scope->__pyx_v_alignment);
                 __pyx_t_15 = 0;
-                __pyx_t_15 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_Rule)), ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1111; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_15 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_Rule)), ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1110; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_15);
                 __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
                 __pyx_r = __pyx_t_15;
@@ -48258,7 +48310,7 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
                 __pyx_t_22 = __pyx_cur_scope->__pyx_t_7;
                 __pyx_t_23 = __pyx_cur_scope->__pyx_t_8;
                 __pyx_t_24 = __pyx_cur_scope->__pyx_t_9;
-                if (unlikely(!__pyx_sent_value)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1111; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                if (unlikely(!__pyx_sent_value)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1110; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               }
               __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
             }
@@ -48273,38 +48325,38 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
       }
       __pyx_L32:;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1113
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1112
  *                                     yield Rule(self.category, f, e, scores, alignment)
  * 
  *                 if len(phrase) < self.max_length and i+spanlen < len(fwords) and pathlen+1 <= self.max_initial_size:             # <<<<<<<<<<<<<<
  *                     for alt_id in range(len(fwords[i+spanlen])):
  *                         new_frontier.append((k, i+spanlen, alt_id, pathlen + 1, node, phrase, is_shadow_path))
  */
-      __pyx_t_22 = PyObject_Length(__pyx_cur_scope->__pyx_v_phrase); if (unlikely(__pyx_t_22 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1113; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_22 = PyObject_Length(__pyx_cur_scope->__pyx_v_phrase); if (unlikely(__pyx_t_22 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1112; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __pyx_t_19 = (__pyx_t_22 < __pyx_cur_scope->__pyx_v_self->max_length);
       if (__pyx_t_19) {
-        __pyx_t_10 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_i); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1113; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_10 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_i); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1112; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_10);
-        __pyx_t_3 = PyNumber_Add(__pyx_t_10, __pyx_cur_scope->__pyx_v_spanlen); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1113; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_3 = PyNumber_Add(__pyx_t_10, __pyx_cur_scope->__pyx_v_spanlen); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1112; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_3);
         __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
-        __pyx_t_22 = PyObject_Length(__pyx_cur_scope->__pyx_v_fwords); if (unlikely(__pyx_t_22 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1113; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-        __pyx_t_10 = PyInt_FromSsize_t(__pyx_t_22); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1113; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_22 = PyObject_Length(__pyx_cur_scope->__pyx_v_fwords); if (unlikely(__pyx_t_22 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1112; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_10 = PyInt_FromSsize_t(__pyx_t_22); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1112; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_10);
-        __pyx_t_15 = PyObject_RichCompare(__pyx_t_3, __pyx_t_10, Py_LT); __Pyx_XGOTREF(__pyx_t_15); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1113; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_15 = PyObject_RichCompare(__pyx_t_3, __pyx_t_10, Py_LT); __Pyx_XGOTREF(__pyx_t_15); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1112; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
         __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
-        __pyx_t_8 = __Pyx_PyObject_IsTrue(__pyx_t_15); if (unlikely(__pyx_t_8 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1113; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_8 = __Pyx_PyObject_IsTrue(__pyx_t_15); if (unlikely(__pyx_t_8 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1112; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
         if (__pyx_t_8) {
-          __pyx_t_15 = PyNumber_Add(__pyx_cur_scope->__pyx_v_pathlen, __pyx_int_1); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1113; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_15 = PyNumber_Add(__pyx_cur_scope->__pyx_v_pathlen, __pyx_int_1); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1112; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_15);
-          __pyx_t_10 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_self->max_initial_size); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1113; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_10 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_self->max_initial_size); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1112; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_10);
-          __pyx_t_3 = PyObject_RichCompare(__pyx_t_15, __pyx_t_10, Py_LE); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1113; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_3 = PyObject_RichCompare(__pyx_t_15, __pyx_t_10, Py_LE); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1112; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
           __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
-          __pyx_t_26 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_26 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1113; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_26 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_26 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1112; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
           __pyx_t_27 = __pyx_t_26;
         } else {
@@ -48316,45 +48368,45 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
       }
       if (__pyx_t_8) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1114
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1113
  * 
  *                 if len(phrase) < self.max_length and i+spanlen < len(fwords) and pathlen+1 <= self.max_initial_size:
  *                     for alt_id in range(len(fwords[i+spanlen])):             # <<<<<<<<<<<<<<
  *                         new_frontier.append((k, i+spanlen, alt_id, pathlen + 1, node, phrase, is_shadow_path))
  *                     num_subpatterns = arity
  */
-        __pyx_t_3 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_i); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1114; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_3 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_i); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1113; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_3);
-        __pyx_t_10 = PyNumber_Add(__pyx_t_3, __pyx_cur_scope->__pyx_v_spanlen); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1114; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_10 = PyNumber_Add(__pyx_t_3, __pyx_cur_scope->__pyx_v_spanlen); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1113; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_10);
         __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-        __pyx_t_3 = PyObject_GetItem(__pyx_cur_scope->__pyx_v_fwords, __pyx_t_10); if (!__pyx_t_3) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1114; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_3 = PyObject_GetItem(__pyx_cur_scope->__pyx_v_fwords, __pyx_t_10); if (!__pyx_t_3) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1113; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_3);
         __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
-        __pyx_t_22 = PyObject_Length(__pyx_t_3); if (unlikely(__pyx_t_22 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1114; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_22 = PyObject_Length(__pyx_t_3); if (unlikely(__pyx_t_22 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1113; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
         for (__pyx_t_18 = 0; __pyx_t_18 < __pyx_t_22; __pyx_t_18+=1) {
           __pyx_cur_scope->__pyx_v_alt_id = __pyx_t_18;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1115
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1114
  *                 if len(phrase) < self.max_length and i+spanlen < len(fwords) and pathlen+1 <= self.max_initial_size:
  *                     for alt_id in range(len(fwords[i+spanlen])):
  *                         new_frontier.append((k, i+spanlen, alt_id, pathlen + 1, node, phrase, is_shadow_path))             # <<<<<<<<<<<<<<
  *                     num_subpatterns = arity
  *                     if not is_shadow_path:
  */
-          __pyx_t_3 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_k); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1115; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_3 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_k); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1114; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_3);
-          __pyx_t_10 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_i); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1115; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_10 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_i); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1114; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_10);
-          __pyx_t_15 = PyNumber_Add(__pyx_t_10, __pyx_cur_scope->__pyx_v_spanlen); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1115; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_15 = PyNumber_Add(__pyx_t_10, __pyx_cur_scope->__pyx_v_spanlen); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1114; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_15);
           __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
-          __pyx_t_10 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_alt_id); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1115; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_10 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_alt_id); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1114; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_10);
-          __pyx_t_2 = PyNumber_Add(__pyx_cur_scope->__pyx_v_pathlen, __pyx_int_1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1115; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_2 = PyNumber_Add(__pyx_cur_scope->__pyx_v_pathlen, __pyx_int_1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1114; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_2);
-          __pyx_t_9 = PyTuple_New(7); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1115; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_9 = PyTuple_New(7); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1114; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_9);
           PyTuple_SET_ITEM(__pyx_t_9, 0, __pyx_t_3);
           __Pyx_GIVEREF(__pyx_t_3);
@@ -48377,11 +48429,11 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
           __pyx_t_15 = 0;
           __pyx_t_10 = 0;
           __pyx_t_2 = 0;
-          __pyx_t_11 = PyList_Append(__pyx_cur_scope->__pyx_v_new_frontier, ((PyObject *)__pyx_t_9)); if (unlikely(__pyx_t_11 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1115; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_11 = PyList_Append(__pyx_cur_scope->__pyx_v_new_frontier, ((PyObject *)__pyx_t_9)); if (unlikely(__pyx_t_11 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1114; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_DECREF(((PyObject *)__pyx_t_9)); __pyx_t_9 = 0;
         }
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1116
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1115
  *                     for alt_id in range(len(fwords[i+spanlen])):
  *                         new_frontier.append((k, i+spanlen, alt_id, pathlen + 1, node, phrase, is_shadow_path))
  *                     num_subpatterns = arity             # <<<<<<<<<<<<<<
@@ -48390,18 +48442,18 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
  */
         __pyx_cur_scope->__pyx_v_num_subpatterns = __pyx_cur_scope->__pyx_v_arity;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1117
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1116
  *                         new_frontier.append((k, i+spanlen, alt_id, pathlen + 1, node, phrase, is_shadow_path))
  *                     num_subpatterns = arity
  *                     if not is_shadow_path:             # <<<<<<<<<<<<<<
  *                         num_subpatterns = num_subpatterns + 1
  *                     if len(phrase)+1 < self.max_length and arity < self.max_nonterminals and num_subpatterns < self.max_chunks:
  */
-        __pyx_t_8 = __Pyx_PyObject_IsTrue(__pyx_cur_scope->__pyx_v_is_shadow_path); if (unlikely(__pyx_t_8 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1117; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_8 = __Pyx_PyObject_IsTrue(__pyx_cur_scope->__pyx_v_is_shadow_path); if (unlikely(__pyx_t_8 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1116; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __pyx_t_19 = (!__pyx_t_8);
         if (__pyx_t_19) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1118
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1117
  *                     num_subpatterns = arity
  *                     if not is_shadow_path:
  *                         num_subpatterns = num_subpatterns + 1             # <<<<<<<<<<<<<<
@@ -48413,14 +48465,14 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
         }
         __pyx_L64:;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1119
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1118
  *                     if not is_shadow_path:
  *                         num_subpatterns = num_subpatterns + 1
  *                     if len(phrase)+1 < self.max_length and arity < self.max_nonterminals and num_subpatterns < self.max_chunks:             # <<<<<<<<<<<<<<
  *                         xcat = sym_setindex(self.category, arity+1)
  *                         xnode = node.children[xcat]
  */
-        __pyx_t_22 = PyObject_Length(__pyx_cur_scope->__pyx_v_phrase); if (unlikely(__pyx_t_22 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1119; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_22 = PyObject_Length(__pyx_cur_scope->__pyx_v_phrase); if (unlikely(__pyx_t_22 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1118; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __pyx_t_19 = ((__pyx_t_22 + 1) < __pyx_cur_scope->__pyx_v_self->max_length);
         if (__pyx_t_19) {
           __pyx_t_8 = (__pyx_cur_scope->__pyx_v_arity < __pyx_cur_scope->__pyx_v_self->max_nonterminals);
@@ -48436,7 +48488,7 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
         }
         if (__pyx_t_8) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1120
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1119
  *                         num_subpatterns = num_subpatterns + 1
  *                     if len(phrase)+1 < self.max_length and arity < self.max_nonterminals and num_subpatterns < self.max_chunks:
  *                         xcat = sym_setindex(self.category, arity+1)             # <<<<<<<<<<<<<<
@@ -48445,16 +48497,16 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
  */
           __pyx_cur_scope->__pyx_v_xcat = __pyx_f_3_sa_sym_setindex(__pyx_cur_scope->__pyx_v_self->category, (__pyx_cur_scope->__pyx_v_arity + 1));
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1121
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1120
  *                     if len(phrase)+1 < self.max_length and arity < self.max_nonterminals and num_subpatterns < self.max_chunks:
  *                         xcat = sym_setindex(self.category, arity+1)
  *                         xnode = node.children[xcat]             # <<<<<<<<<<<<<<
  *                         # I put spanlen=1 below
  *                         key = tuple([self.min_gap_size, i, 1, pathlen])
  */
-          __pyx_t_9 = PyObject_GetAttr(__pyx_cur_scope->__pyx_v_node, __pyx_n_s__children); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1121; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_9 = PyObject_GetAttr(__pyx_cur_scope->__pyx_v_node, __pyx_n_s__children); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1120; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_9);
-          __pyx_t_2 = __Pyx_GetItemInt(__pyx_t_9, __pyx_cur_scope->__pyx_v_xcat, sizeof(int), PyInt_FromLong); if (!__pyx_t_2) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1121; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_2 = __Pyx_GetItemInt(__pyx_t_9, __pyx_cur_scope->__pyx_v_xcat, sizeof(int), PyInt_FromLong); if (!__pyx_t_2) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1120; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_2);
           __Pyx_DECREF(__pyx_t_9); __pyx_t_9 = 0;
           __Pyx_XGOTREF(__pyx_cur_scope->__pyx_v_xnode);
@@ -48463,18 +48515,18 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
           __pyx_cur_scope->__pyx_v_xnode = __pyx_t_2;
           __pyx_t_2 = 0;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1123
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1122
  *                         xnode = node.children[xcat]
  *                         # I put spanlen=1 below
  *                         key = tuple([self.min_gap_size, i, 1, pathlen])             # <<<<<<<<<<<<<<
  *                         frontier_nodes = []
  *                         if (key in nodes_isteps_away_buffer):
  */
-          __pyx_t_2 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_self->min_gap_size); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1123; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_2 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_self->min_gap_size); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1122; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_2);
-          __pyx_t_9 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_i); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1123; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_9 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_i); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1122; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_9);
-          __pyx_t_10 = PyList_New(4); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1123; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_10 = PyList_New(4); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1122; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_10);
           PyList_SET_ITEM(__pyx_t_10, 0, __pyx_t_2);
           __Pyx_GIVEREF(__pyx_t_2);
@@ -48488,7 +48540,7 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
           __Pyx_GIVEREF(__pyx_cur_scope->__pyx_v_pathlen);
           __pyx_t_2 = 0;
           __pyx_t_9 = 0;
-          __pyx_t_9 = ((PyObject *)PyList_AsTuple(__pyx_t_10)); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1123; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_9 = ((PyObject *)PyList_AsTuple(__pyx_t_10)); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1122; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(((PyObject *)__pyx_t_9));
           __Pyx_DECREF(((PyObject *)__pyx_t_10)); __pyx_t_10 = 0;
           __Pyx_XGOTREF(((PyObject *)__pyx_cur_scope->__pyx_v_key));
@@ -48497,14 +48549,14 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
           __pyx_cur_scope->__pyx_v_key = __pyx_t_9;
           __pyx_t_9 = 0;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1124
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1123
  *                         # I put spanlen=1 below
  *                         key = tuple([self.min_gap_size, i, 1, pathlen])
  *                         frontier_nodes = []             # <<<<<<<<<<<<<<
  *                         if (key in nodes_isteps_away_buffer):
  *                             frontier_nodes = nodes_isteps_away_buffer[key]
  */
-          __pyx_t_9 = PyList_New(0); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1124; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_9 = PyList_New(0); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1123; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_9);
           __Pyx_XGOTREF(__pyx_cur_scope->__pyx_v_frontier_nodes);
           __Pyx_XDECREF(__pyx_cur_scope->__pyx_v_frontier_nodes);
@@ -48512,24 +48564,24 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
           __pyx_cur_scope->__pyx_v_frontier_nodes = ((PyObject *)__pyx_t_9);
           __pyx_t_9 = 0;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1125
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1124
  *                         key = tuple([self.min_gap_size, i, 1, pathlen])
  *                         frontier_nodes = []
  *                         if (key in nodes_isteps_away_buffer):             # <<<<<<<<<<<<<<
  *                             frontier_nodes = nodes_isteps_away_buffer[key]
  *                         else:
  */
-          __pyx_t_8 = (__Pyx_PyDict_Contains(((PyObject *)__pyx_cur_scope->__pyx_v_key), ((PyObject *)__pyx_cur_scope->__pyx_v_nodes_isteps_away_buffer), Py_EQ)); if (unlikely(__pyx_t_8 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1125; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_8 = (__Pyx_PyDict_Contains(((PyObject *)__pyx_cur_scope->__pyx_v_key), ((PyObject *)__pyx_cur_scope->__pyx_v_nodes_isteps_away_buffer), Py_EQ)); if (unlikely(__pyx_t_8 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1124; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           if (__pyx_t_8) {
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1126
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1125
  *                         frontier_nodes = []
  *                         if (key in nodes_isteps_away_buffer):
  *                             frontier_nodes = nodes_isteps_away_buffer[key]             # <<<<<<<<<<<<<<
  *                         else:
  *                             frontier_nodes = self.get_all_nodes_isteps_away(self.min_gap_size, i, 1, pathlen, fwords, next_states, reachable_buffer)
  */
-            __pyx_t_9 = __Pyx_PyDict_GetItem(((PyObject *)__pyx_cur_scope->__pyx_v_nodes_isteps_away_buffer), ((PyObject *)__pyx_cur_scope->__pyx_v_key)); if (!__pyx_t_9) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1126; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_9 = __Pyx_PyDict_GetItem(((PyObject *)__pyx_cur_scope->__pyx_v_nodes_isteps_away_buffer), ((PyObject *)__pyx_cur_scope->__pyx_v_key)); if (!__pyx_t_9) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1125; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_9);
             __Pyx_GOTREF(__pyx_cur_scope->__pyx_v_frontier_nodes);
             __Pyx_DECREF(__pyx_cur_scope->__pyx_v_frontier_nodes);
@@ -48540,20 +48592,20 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
           }
           /*else*/ {
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1128
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1127
  *                             frontier_nodes = nodes_isteps_away_buffer[key]
  *                         else:
  *                             frontier_nodes = self.get_all_nodes_isteps_away(self.min_gap_size, i, 1, pathlen, fwords, next_states, reachable_buffer)             # <<<<<<<<<<<<<<
  *                             nodes_isteps_away_buffer[key] = frontier_nodes
  * 
  */
-            __pyx_t_9 = PyObject_GetAttr(((PyObject *)__pyx_cur_scope->__pyx_v_self), __pyx_n_s_124); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1128; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_9 = PyObject_GetAttr(((PyObject *)__pyx_cur_scope->__pyx_v_self), __pyx_n_s_124); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1127; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_9);
-            __pyx_t_10 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_self->min_gap_size); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1128; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_10 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_self->min_gap_size); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1127; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_10);
-            __pyx_t_2 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_i); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1128; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_2 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_i); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1127; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_2);
-            __pyx_t_15 = PyTuple_New(7); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1128; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_15 = PyTuple_New(7); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1127; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_15);
             PyTuple_SET_ITEM(__pyx_t_15, 0, __pyx_t_10);
             __Pyx_GIVEREF(__pyx_t_10);
@@ -48576,7 +48628,7 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
             __Pyx_GIVEREF(((PyObject *)__pyx_cur_scope->__pyx_v_reachable_buffer));
             __pyx_t_10 = 0;
             __pyx_t_2 = 0;
-            __pyx_t_2 = PyObject_Call(__pyx_t_9, ((PyObject *)__pyx_t_15), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1128; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_2 = PyObject_Call(__pyx_t_9, ((PyObject *)__pyx_t_15), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1127; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_2);
             __Pyx_DECREF(__pyx_t_9); __pyx_t_9 = 0;
             __Pyx_DECREF(((PyObject *)__pyx_t_15)); __pyx_t_15 = 0;
@@ -48586,18 +48638,18 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
             __pyx_cur_scope->__pyx_v_frontier_nodes = __pyx_t_2;
             __pyx_t_2 = 0;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1129
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1128
  *                         else:
  *                             frontier_nodes = self.get_all_nodes_isteps_away(self.min_gap_size, i, 1, pathlen, fwords, next_states, reachable_buffer)
  *                             nodes_isteps_away_buffer[key] = frontier_nodes             # <<<<<<<<<<<<<<
  * 
  *                         for (i, alt, pathlen) in frontier_nodes:
  */
-            if (PyDict_SetItem(((PyObject *)__pyx_cur_scope->__pyx_v_nodes_isteps_away_buffer), ((PyObject *)__pyx_cur_scope->__pyx_v_key), __pyx_cur_scope->__pyx_v_frontier_nodes) < 0) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1129; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            if (PyDict_SetItem(((PyObject *)__pyx_cur_scope->__pyx_v_nodes_isteps_away_buffer), ((PyObject *)__pyx_cur_scope->__pyx_v_key), __pyx_cur_scope->__pyx_v_frontier_nodes) < 0) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1128; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           }
           __pyx_L66:;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1131
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1130
  *                             nodes_isteps_away_buffer[key] = frontier_nodes
  * 
  *                         for (i, alt, pathlen) in frontier_nodes:             # <<<<<<<<<<<<<<
@@ -48608,7 +48660,7 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
             __pyx_t_2 = __pyx_cur_scope->__pyx_v_frontier_nodes; __Pyx_INCREF(__pyx_t_2); __pyx_t_22 = 0;
             __pyx_t_20 = NULL;
           } else {
-            __pyx_t_22 = -1; __pyx_t_2 = PyObject_GetIter(__pyx_cur_scope->__pyx_v_frontier_nodes); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_22 = -1; __pyx_t_2 = PyObject_GetIter(__pyx_cur_scope->__pyx_v_frontier_nodes); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_2);
             __pyx_t_20 = Py_TYPE(__pyx_t_2)->tp_iternext;
           }
@@ -48616,23 +48668,23 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
             if (!__pyx_t_20 && PyList_CheckExact(__pyx_t_2)) {
               if (__pyx_t_22 >= PyList_GET_SIZE(__pyx_t_2)) break;
               #if CYTHON_COMPILING_IN_CPYTHON
-              __pyx_t_15 = PyList_GET_ITEM(__pyx_t_2, __pyx_t_22); __Pyx_INCREF(__pyx_t_15); __pyx_t_22++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_15 = PyList_GET_ITEM(__pyx_t_2, __pyx_t_22); __Pyx_INCREF(__pyx_t_15); __pyx_t_22++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               #else
-              __pyx_t_15 = PySequence_ITEM(__pyx_t_2, __pyx_t_22); __pyx_t_22++; if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_15 = PySequence_ITEM(__pyx_t_2, __pyx_t_22); __pyx_t_22++; if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               #endif
             } else if (!__pyx_t_20 && PyTuple_CheckExact(__pyx_t_2)) {
               if (__pyx_t_22 >= PyTuple_GET_SIZE(__pyx_t_2)) break;
               #if CYTHON_COMPILING_IN_CPYTHON
-              __pyx_t_15 = PyTuple_GET_ITEM(__pyx_t_2, __pyx_t_22); __Pyx_INCREF(__pyx_t_15); __pyx_t_22++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_15 = PyTuple_GET_ITEM(__pyx_t_2, __pyx_t_22); __Pyx_INCREF(__pyx_t_15); __pyx_t_22++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               #else
-              __pyx_t_15 = PySequence_ITEM(__pyx_t_2, __pyx_t_22); __pyx_t_22++; if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_15 = PySequence_ITEM(__pyx_t_2, __pyx_t_22); __pyx_t_22++; if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               #endif
             } else {
               __pyx_t_15 = __pyx_t_20(__pyx_t_2);
               if (unlikely(!__pyx_t_15)) {
                 if (PyErr_Occurred()) {
                   if (likely(PyErr_ExceptionMatches(PyExc_StopIteration))) PyErr_Clear();
-                  else {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  else {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 }
                 break;
               }
@@ -48648,7 +48700,7 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
               if (unlikely(size != 3)) {
                 if (size > 3) __Pyx_RaiseTooManyValuesError(3);
                 else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size);
-                {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               }
               #if CYTHON_COMPILING_IN_CPYTHON
               if (likely(PyTuple_CheckExact(sequence))) {
@@ -48664,38 +48716,38 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
               __Pyx_INCREF(__pyx_t_10);
               __Pyx_INCREF(__pyx_t_3);
               #else
-              __pyx_t_9 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-              __pyx_t_10 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-              __pyx_t_3 = PySequence_ITEM(sequence, 2); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_9 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_10 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_3 = PySequence_ITEM(sequence, 2); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               #endif
               __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
             } else
             {
               Py_ssize_t index = -1;
-              __pyx_t_14 = PyObject_GetIter(__pyx_t_15); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-              __Pyx_GOTREF(__pyx_t_14);
+              __pyx_t_13 = PyObject_GetIter(__pyx_t_15); if (unlikely(!__pyx_t_13)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __Pyx_GOTREF(__pyx_t_13);
               __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
-              __pyx_t_17 = Py_TYPE(__pyx_t_14)->tp_iternext;
-              index = 0; __pyx_t_9 = __pyx_t_17(__pyx_t_14); if (unlikely(!__pyx_t_9)) goto __pyx_L69_unpacking_failed;
+              __pyx_t_17 = Py_TYPE(__pyx_t_13)->tp_iternext;
+              index = 0; __pyx_t_9 = __pyx_t_17(__pyx_t_13); if (unlikely(!__pyx_t_9)) goto __pyx_L69_unpacking_failed;
               __Pyx_GOTREF(__pyx_t_9);
-              index = 1; __pyx_t_10 = __pyx_t_17(__pyx_t_14); if (unlikely(!__pyx_t_10)) goto __pyx_L69_unpacking_failed;
+              index = 1; __pyx_t_10 = __pyx_t_17(__pyx_t_13); if (unlikely(!__pyx_t_10)) goto __pyx_L69_unpacking_failed;
               __Pyx_GOTREF(__pyx_t_10);
-              index = 2; __pyx_t_3 = __pyx_t_17(__pyx_t_14); if (unlikely(!__pyx_t_3)) goto __pyx_L69_unpacking_failed;
+              index = 2; __pyx_t_3 = __pyx_t_17(__pyx_t_13); if (unlikely(!__pyx_t_3)) goto __pyx_L69_unpacking_failed;
               __Pyx_GOTREF(__pyx_t_3);
-              if (__Pyx_IternextUnpackEndCheck(__pyx_t_17(__pyx_t_14), 3) < 0) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              if (__Pyx_IternextUnpackEndCheck(__pyx_t_17(__pyx_t_13), 3) < 0) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __pyx_t_17 = NULL;
-              __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
+              __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0;
               goto __pyx_L70_unpacking_done;
               __pyx_L69_unpacking_failed:;
-              __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
+              __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0;
               __pyx_t_17 = NULL;
               if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index);
-              {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __pyx_L70_unpacking_done:;
             }
-            __pyx_t_18 = __Pyx_PyInt_AsInt(__pyx_t_9); if (unlikely((__pyx_t_18 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_18 = __Pyx_PyInt_AsInt(__pyx_t_9); if (unlikely((__pyx_t_18 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_DECREF(__pyx_t_9); __pyx_t_9 = 0;
-            __pyx_t_6 = __Pyx_PyInt_AsInt(__pyx_t_10); if (unlikely((__pyx_t_6 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_6 = __Pyx_PyInt_AsInt(__pyx_t_10); if (unlikely((__pyx_t_6 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
             __pyx_cur_scope->__pyx_v_i = __pyx_t_18;
             __pyx_cur_scope->__pyx_v_alt = __pyx_t_6;
@@ -48705,54 +48757,54 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
             __pyx_cur_scope->__pyx_v_pathlen = __pyx_t_3;
             __pyx_t_3 = 0;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1132
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1131
  * 
  *                         for (i, alt, pathlen) in frontier_nodes:
  *                             new_frontier.append((k, i, alt, pathlen, xnode, phrase +(xcat,), is_shadow_path))             # <<<<<<<<<<<<<<
  *             frontier = new_frontier
  * 
  */
-            __pyx_t_15 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_k); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_15 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_k); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_15);
-            __pyx_t_3 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_i); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_3 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_i); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_3);
-            __pyx_t_10 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_alt); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_10 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_alt); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_10);
-            __pyx_t_9 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_xcat); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_9 = PyInt_FromLong(__pyx_cur_scope->__pyx_v_xcat); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_9);
-            __pyx_t_14 = PyTuple_New(1); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-            __Pyx_GOTREF(__pyx_t_14);
-            PyTuple_SET_ITEM(__pyx_t_14, 0, __pyx_t_9);
+            __pyx_t_13 = PyTuple_New(1); if (unlikely(!__pyx_t_13)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __Pyx_GOTREF(__pyx_t_13);
+            PyTuple_SET_ITEM(__pyx_t_13, 0, __pyx_t_9);
             __Pyx_GIVEREF(__pyx_t_9);
             __pyx_t_9 = 0;
-            __pyx_t_9 = PyNumber_Add(__pyx_cur_scope->__pyx_v_phrase, ((PyObject *)__pyx_t_14)); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_9 = PyNumber_Add(__pyx_cur_scope->__pyx_v_phrase, ((PyObject *)__pyx_t_13)); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_9);
-            __Pyx_DECREF(((PyObject *)__pyx_t_14)); __pyx_t_14 = 0;
-            __pyx_t_14 = PyTuple_New(7); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-            __Pyx_GOTREF(__pyx_t_14);
-            PyTuple_SET_ITEM(__pyx_t_14, 0, __pyx_t_15);
+            __Pyx_DECREF(((PyObject *)__pyx_t_13)); __pyx_t_13 = 0;
+            __pyx_t_13 = PyTuple_New(7); if (unlikely(!__pyx_t_13)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __Pyx_GOTREF(__pyx_t_13);
+            PyTuple_SET_ITEM(__pyx_t_13, 0, __pyx_t_15);
             __Pyx_GIVEREF(__pyx_t_15);
-            PyTuple_SET_ITEM(__pyx_t_14, 1, __pyx_t_3);
+            PyTuple_SET_ITEM(__pyx_t_13, 1, __pyx_t_3);
             __Pyx_GIVEREF(__pyx_t_3);
-            PyTuple_SET_ITEM(__pyx_t_14, 2, __pyx_t_10);
+            PyTuple_SET_ITEM(__pyx_t_13, 2, __pyx_t_10);
             __Pyx_GIVEREF(__pyx_t_10);
             __Pyx_INCREF(__pyx_cur_scope->__pyx_v_pathlen);
-            PyTuple_SET_ITEM(__pyx_t_14, 3, __pyx_cur_scope->__pyx_v_pathlen);
+            PyTuple_SET_ITEM(__pyx_t_13, 3, __pyx_cur_scope->__pyx_v_pathlen);
             __Pyx_GIVEREF(__pyx_cur_scope->__pyx_v_pathlen);
             __Pyx_INCREF(__pyx_cur_scope->__pyx_v_xnode);
-            PyTuple_SET_ITEM(__pyx_t_14, 4, __pyx_cur_scope->__pyx_v_xnode);
+            PyTuple_SET_ITEM(__pyx_t_13, 4, __pyx_cur_scope->__pyx_v_xnode);
             __Pyx_GIVEREF(__pyx_cur_scope->__pyx_v_xnode);
-            PyTuple_SET_ITEM(__pyx_t_14, 5, __pyx_t_9);
+            PyTuple_SET_ITEM(__pyx_t_13, 5, __pyx_t_9);
             __Pyx_GIVEREF(__pyx_t_9);
             __Pyx_INCREF(__pyx_cur_scope->__pyx_v_is_shadow_path);
-            PyTuple_SET_ITEM(__pyx_t_14, 6, __pyx_cur_scope->__pyx_v_is_shadow_path);
+            PyTuple_SET_ITEM(__pyx_t_13, 6, __pyx_cur_scope->__pyx_v_is_shadow_path);
             __Pyx_GIVEREF(__pyx_cur_scope->__pyx_v_is_shadow_path);
             __pyx_t_15 = 0;
             __pyx_t_3 = 0;
             __pyx_t_10 = 0;
             __pyx_t_9 = 0;
-            __pyx_t_11 = PyList_Append(__pyx_cur_scope->__pyx_v_new_frontier, ((PyObject *)__pyx_t_14)); if (unlikely(__pyx_t_11 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-            __Pyx_DECREF(((PyObject *)__pyx_t_14)); __pyx_t_14 = 0;
+            __pyx_t_11 = PyList_Append(__pyx_cur_scope->__pyx_v_new_frontier, ((PyObject *)__pyx_t_13)); if (unlikely(__pyx_t_11 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __Pyx_DECREF(((PyObject *)__pyx_t_13)); __pyx_t_13 = 0;
           }
           __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
           goto __pyx_L65;
@@ -48765,7 +48817,7 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
     }
     __Pyx_DECREF(__pyx_t_12); __pyx_t_12 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1133
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1132
  *                         for (i, alt, pathlen) in frontier_nodes:
  *                             new_frontier.append((k, i, alt, pathlen, xnode, phrase +(xcat,), is_shadow_path))
  *             frontier = new_frontier             # <<<<<<<<<<<<<<
@@ -48779,94 +48831,94 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
     __pyx_cur_scope->__pyx_v_frontier = __pyx_cur_scope->__pyx_v_new_frontier;
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1135
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1134
  *             frontier = new_frontier
  * 
  *         stop_time = monitor_cpu()             # <<<<<<<<<<<<<<
  *         logger.info("Total time for rule lookup, extraction, and scoring = %f seconds", (stop_time - start_time))
  *         gc.collect()
  */
-  __pyx_t_12 = PyFloat_FromDouble(__pyx_f_3_sa_monitor_cpu()); if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1135; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_12 = PyFloat_FromDouble(__pyx_f_3_sa_monitor_cpu()); if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1134; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_12);
   __Pyx_GIVEREF(__pyx_t_12);
   __pyx_cur_scope->__pyx_v_stop_time = __pyx_t_12;
   __pyx_t_12 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1136
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1135
  * 
  *         stop_time = monitor_cpu()
  *         logger.info("Total time for rule lookup, extraction, and scoring = %f seconds", (stop_time - start_time))             # <<<<<<<<<<<<<<
  *         gc.collect()
  *         logger.info("    Extract time = %f seconds", self.extract_time)
  */
-  __pyx_t_12 = __Pyx_GetName(__pyx_m, __pyx_n_s__logger); if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1136; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_12 = __Pyx_GetName(__pyx_m, __pyx_n_s__logger); if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1135; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_12);
-  __pyx_t_2 = PyObject_GetAttr(__pyx_t_12, __pyx_n_s__info); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1136; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_t_12, __pyx_n_s__info); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1135; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_12); __pyx_t_12 = 0;
-  __pyx_t_12 = PyFloat_FromDouble(__pyx_cur_scope->__pyx_v_start_time); if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1136; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_12 = PyFloat_FromDouble(__pyx_cur_scope->__pyx_v_start_time); if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1135; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_12);
-  __pyx_t_14 = PyNumber_Subtract(__pyx_cur_scope->__pyx_v_stop_time, __pyx_t_12); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1136; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_14);
+  __pyx_t_13 = PyNumber_Subtract(__pyx_cur_scope->__pyx_v_stop_time, __pyx_t_12); if (unlikely(!__pyx_t_13)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1135; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_13);
   __Pyx_DECREF(__pyx_t_12); __pyx_t_12 = 0;
-  __pyx_t_12 = PyTuple_New(2); if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1136; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_12 = PyTuple_New(2); if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1135; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_12);
   __Pyx_INCREF(((PyObject *)__pyx_kp_s_125));
   PyTuple_SET_ITEM(__pyx_t_12, 0, ((PyObject *)__pyx_kp_s_125));
   __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_125));
-  PyTuple_SET_ITEM(__pyx_t_12, 1, __pyx_t_14);
-  __Pyx_GIVEREF(__pyx_t_14);
-  __pyx_t_14 = 0;
-  __pyx_t_14 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_12), NULL); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1136; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_14);
+  PyTuple_SET_ITEM(__pyx_t_12, 1, __pyx_t_13);
+  __Pyx_GIVEREF(__pyx_t_13);
+  __pyx_t_13 = 0;
+  __pyx_t_13 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_12), NULL); if (unlikely(!__pyx_t_13)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1135; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_13);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_12)); __pyx_t_12 = 0;
-  __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
+  __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1137
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1136
  *         stop_time = monitor_cpu()
  *         logger.info("Total time for rule lookup, extraction, and scoring = %f seconds", (stop_time - start_time))
  *         gc.collect()             # <<<<<<<<<<<<<<
  *         logger.info("    Extract time = %f seconds", self.extract_time)
  * 
  */
-  __pyx_t_14 = __Pyx_GetName(__pyx_m, __pyx_n_s__gc); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_14);
-  __pyx_t_12 = PyObject_GetAttr(__pyx_t_14, __pyx_n_s__collect); if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_13 = __Pyx_GetName(__pyx_m, __pyx_n_s__gc); if (unlikely(!__pyx_t_13)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1136; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_13);
+  __pyx_t_12 = PyObject_GetAttr(__pyx_t_13, __pyx_n_s__collect); if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1136; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_12);
-  __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
-  __pyx_t_14 = PyObject_Call(__pyx_t_12, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_14);
+  __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0;
+  __pyx_t_13 = PyObject_Call(__pyx_t_12, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_13)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1136; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_13);
   __Pyx_DECREF(__pyx_t_12); __pyx_t_12 = 0;
-  __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
+  __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1138
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1137
  *         logger.info("Total time for rule lookup, extraction, and scoring = %f seconds", (stop_time - start_time))
  *         gc.collect()
  *         logger.info("    Extract time = %f seconds", self.extract_time)             # <<<<<<<<<<<<<<
  * 
  * 
  */
-  __pyx_t_14 = __Pyx_GetName(__pyx_m, __pyx_n_s__logger); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1138; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_14);
-  __pyx_t_12 = PyObject_GetAttr(__pyx_t_14, __pyx_n_s__info); if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1138; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_13 = __Pyx_GetName(__pyx_m, __pyx_n_s__logger); if (unlikely(!__pyx_t_13)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_13);
+  __pyx_t_12 = PyObject_GetAttr(__pyx_t_13, __pyx_n_s__info); if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_12);
-  __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
-  __pyx_t_14 = PyFloat_FromDouble(__pyx_cur_scope->__pyx_v_self->extract_time); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1138; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_14);
-  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1138; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0;
+  __pyx_t_13 = PyFloat_FromDouble(__pyx_cur_scope->__pyx_v_self->extract_time); if (unlikely(!__pyx_t_13)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_13);
+  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)__pyx_kp_s_126));
   PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_kp_s_126));
   __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_126));
-  PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_t_14);
-  __Pyx_GIVEREF(__pyx_t_14);
-  __pyx_t_14 = 0;
-  __pyx_t_14 = PyObject_Call(__pyx_t_12, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1138; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_14);
+  PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_t_13);
+  __Pyx_GIVEREF(__pyx_t_13);
+  __pyx_t_13 = 0;
+  __pyx_t_13 = PyObject_Call(__pyx_t_12, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_13)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_13);
   __Pyx_DECREF(__pyx_t_12); __pyx_t_12 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
-  __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
+  __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0;
   PyErr_SetNone(PyExc_StopIteration);
   goto __pyx_L0;
   __pyx_L1_error:;
@@ -48889,7 +48941,7 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
   return NULL;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1141
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1140
  * 
  * 
  *     cdef int find_fixpoint(self,             # <<<<<<<<<<<<<<
@@ -48919,7 +48971,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("find_fixpoint", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1156
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1155
  *         cdef int e_low_prev, e_high_prev, f_low_prev, f_high_prev, new_x, new_low_x, new_high_x
  * 
  *         e_low[0] = e_in_low             # <<<<<<<<<<<<<<
@@ -48928,7 +48980,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
  */
   (__pyx_v_e_low[0]) = __pyx_v_e_in_low;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1157
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1156
  * 
  *         e_low[0] = e_in_low
  *         e_high[0] = e_in_high             # <<<<<<<<<<<<<<
@@ -48937,19 +48989,19 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
  */
   (__pyx_v_e_high[0]) = __pyx_v_e_in_high;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1158
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1157
  *         e_low[0] = e_in_low
  *         e_high[0] = e_in_high
  *         self.find_projection(f_low, f_high, f_links_low, f_links_high, e_low, e_high)             # <<<<<<<<<<<<<<
  *         if e_low[0] == -1:
  *             # low-priority corner case: if phrase w is unaligned,
  */
-  __pyx_t_1 = __Pyx_PyInt_AsInt(__pyx_v_f_high); if (unlikely((__pyx_t_1 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1158; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __pyx_t_2 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->find_projection(__pyx_v_self, __pyx_v_f_low, __pyx_t_1, __pyx_v_f_links_low, __pyx_v_f_links_high, __pyx_v_e_low, __pyx_v_e_high); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1158; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyInt_AsInt(__pyx_v_f_high); if (unlikely((__pyx_t_1 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1157; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->find_projection(__pyx_v_self, __pyx_v_f_low, __pyx_t_1, __pyx_v_f_links_low, __pyx_v_f_links_high, __pyx_v_e_low, __pyx_v_e_high); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1157; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1159
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1158
  *         e_high[0] = e_in_high
  *         self.find_projection(f_low, f_high, f_links_low, f_links_high, e_low, e_high)
  *         if e_low[0] == -1:             # <<<<<<<<<<<<<<
@@ -48959,7 +49011,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
   __pyx_t_3 = ((__pyx_v_e_low[0]) == -1);
   if (__pyx_t_3) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1165
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1164
  *             # rule X -> X_1 w X_2 / X_1 X_2.    This is probably
  *             # not worth the bother, though.
  *             return 0             # <<<<<<<<<<<<<<
@@ -48971,7 +49023,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     goto __pyx_L3;
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1166
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1165
  *             # not worth the bother, though.
  *             return 0
  *         elif e_in_low != -1 and e_low[0] != e_in_low:             # <<<<<<<<<<<<<<
@@ -48987,7 +49039,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
   }
   if (__pyx_t_5) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1167
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1166
  *             return 0
  *         elif e_in_low != -1 and e_low[0] != e_in_low:
  *             if e_in_low - e_low[0] < min_ex_size:             # <<<<<<<<<<<<<<
@@ -48997,7 +49049,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     __pyx_t_5 = ((__pyx_v_e_in_low - (__pyx_v_e_low[0])) < __pyx_v_min_ex_size);
     if (__pyx_t_5) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1168
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1167
  *         elif e_in_low != -1 and e_low[0] != e_in_low:
  *             if e_in_low - e_low[0] < min_ex_size:
  *                 e_low[0] = e_in_low - min_ex_size             # <<<<<<<<<<<<<<
@@ -49006,7 +49058,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
  */
       (__pyx_v_e_low[0]) = (__pyx_v_e_in_low - __pyx_v_min_ex_size);
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1169
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1168
  *             if e_in_low - e_low[0] < min_ex_size:
  *                 e_low[0] = e_in_low - min_ex_size
  *                 if e_low[0] < 0:             # <<<<<<<<<<<<<<
@@ -49016,7 +49068,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
       __pyx_t_5 = ((__pyx_v_e_low[0]) < 0);
       if (__pyx_t_5) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1170
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1169
  *                 e_low[0] = e_in_low - min_ex_size
  *                 if e_low[0] < 0:
  *                     return 0             # <<<<<<<<<<<<<<
@@ -49035,7 +49087,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
   }
   __pyx_L3:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1172
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1171
  *                     return 0
  * 
  *         if e_high[0] - e_low[0] > max_e_len:             # <<<<<<<<<<<<<<
@@ -49045,7 +49097,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
   __pyx_t_5 = (((__pyx_v_e_high[0]) - (__pyx_v_e_low[0])) > __pyx_v_max_e_len);
   if (__pyx_t_5) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1173
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1172
  * 
  *         if e_high[0] - e_low[0] > max_e_len:
  *             return 0             # <<<<<<<<<<<<<<
@@ -49057,7 +49109,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     goto __pyx_L6;
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1174
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1173
  *         if e_high[0] - e_low[0] > max_e_len:
  *             return 0
  *         elif e_in_high != -1 and e_high[0] != e_in_high:             # <<<<<<<<<<<<<<
@@ -49073,7 +49125,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
   }
   if (__pyx_t_4) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1175
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1174
  *             return 0
  *         elif e_in_high != -1 and e_high[0] != e_in_high:
  *             if e_high[0] - e_in_high < min_ex_size:             # <<<<<<<<<<<<<<
@@ -49083,7 +49135,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     __pyx_t_4 = (((__pyx_v_e_high[0]) - __pyx_v_e_in_high) < __pyx_v_min_ex_size);
     if (__pyx_t_4) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1176
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1175
  *         elif e_in_high != -1 and e_high[0] != e_in_high:
  *             if e_high[0] - e_in_high < min_ex_size:
  *                 e_high[0] = e_in_high + min_ex_size             # <<<<<<<<<<<<<<
@@ -49092,7 +49144,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
  */
       (__pyx_v_e_high[0]) = (__pyx_v_e_in_high + __pyx_v_min_ex_size);
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1177
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1176
  *             if e_high[0] - e_in_high < min_ex_size:
  *                 e_high[0] = e_in_high + min_ex_size
  *                 if e_high[0] > e_sent_len:             # <<<<<<<<<<<<<<
@@ -49102,7 +49154,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
       __pyx_t_4 = ((__pyx_v_e_high[0]) > __pyx_v_e_sent_len);
       if (__pyx_t_4) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1178
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1177
  *                 e_high[0] = e_in_high + min_ex_size
  *                 if e_high[0] > e_sent_len:
  *                     return 0             # <<<<<<<<<<<<<<
@@ -49121,7 +49173,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
   }
   __pyx_L6:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1180
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1179
  *                     return 0
  * 
  *         f_back_low[0] = -1             # <<<<<<<<<<<<<<
@@ -49130,7 +49182,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
  */
   (__pyx_v_f_back_low[0]) = -1;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1181
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1180
  * 
  *         f_back_low[0] = -1
  *         f_back_high[0] = -1             # <<<<<<<<<<<<<<
@@ -49139,7 +49191,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
  */
   (__pyx_v_f_back_high[0]) = -1;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1182
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1181
  *         f_back_low[0] = -1
  *         f_back_high[0] = -1
  *         f_low_prev = f_low             # <<<<<<<<<<<<<<
@@ -49148,17 +49200,17 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
  */
   __pyx_v_f_low_prev = __pyx_v_f_low;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1183
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1182
  *         f_back_high[0] = -1
  *         f_low_prev = f_low
  *         f_high_prev = f_high             # <<<<<<<<<<<<<<
  *         new_x = 0
  *         new_low_x = 0
  */
-  __pyx_t_1 = __Pyx_PyInt_AsInt(__pyx_v_f_high); if (unlikely((__pyx_t_1 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1183; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyInt_AsInt(__pyx_v_f_high); if (unlikely((__pyx_t_1 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1182; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_v_f_high_prev = __pyx_t_1;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1184
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1183
  *         f_low_prev = f_low
  *         f_high_prev = f_high
  *         new_x = 0             # <<<<<<<<<<<<<<
@@ -49167,7 +49219,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
  */
   __pyx_v_new_x = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1185
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1184
  *         f_high_prev = f_high
  *         new_x = 0
  *         new_low_x = 0             # <<<<<<<<<<<<<<
@@ -49176,7 +49228,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
  */
   __pyx_v_new_low_x = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1186
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1185
  *         new_x = 0
  *         new_low_x = 0
  *         new_high_x = 0             # <<<<<<<<<<<<<<
@@ -49185,7 +49237,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
  */
   __pyx_v_new_high_x = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1188
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1187
  *         new_high_x = 0
  * 
  *         while True:             # <<<<<<<<<<<<<<
@@ -49195,7 +49247,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
   while (1) {
     if (!1) break;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1190
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1189
  *         while True:
  * 
  *             if f_back_low[0] == -1:             # <<<<<<<<<<<<<<
@@ -49205,45 +49257,45 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     __pyx_t_4 = ((__pyx_v_f_back_low[0]) == -1);
     if (__pyx_t_4) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1191
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1190
  * 
  *             if f_back_low[0] == -1:
  *                 self.find_projection(e_low[0], e_high[0], e_links_low, e_links_high, f_back_low, f_back_high)             # <<<<<<<<<<<<<<
  *             else:
  *                 self.find_projection(e_low[0], e_low_prev, e_links_low, e_links_high, f_back_low, f_back_high)
  */
-      __pyx_t_2 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->find_projection(__pyx_v_self, (__pyx_v_e_low[0]), (__pyx_v_e_high[0]), __pyx_v_e_links_low, __pyx_v_e_links_high, __pyx_v_f_back_low, __pyx_v_f_back_high); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1191; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->find_projection(__pyx_v_self, (__pyx_v_e_low[0]), (__pyx_v_e_high[0]), __pyx_v_e_links_low, __pyx_v_e_links_high, __pyx_v_f_back_low, __pyx_v_f_back_high); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1190; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
       goto __pyx_L11;
     }
     /*else*/ {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1193
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1192
  *                 self.find_projection(e_low[0], e_high[0], e_links_low, e_links_high, f_back_low, f_back_high)
  *             else:
  *                 self.find_projection(e_low[0], e_low_prev, e_links_low, e_links_high, f_back_low, f_back_high)             # <<<<<<<<<<<<<<
  *                 self.find_projection(e_high_prev, e_high[0], e_links_low, e_links_high, f_back_low, f_back_high)
  * 
  */
-      __pyx_t_2 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->find_projection(__pyx_v_self, (__pyx_v_e_low[0]), __pyx_v_e_low_prev, __pyx_v_e_links_low, __pyx_v_e_links_high, __pyx_v_f_back_low, __pyx_v_f_back_high); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1193; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->find_projection(__pyx_v_self, (__pyx_v_e_low[0]), __pyx_v_e_low_prev, __pyx_v_e_links_low, __pyx_v_e_links_high, __pyx_v_f_back_low, __pyx_v_f_back_high); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1192; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1194
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1193
  *             else:
  *                 self.find_projection(e_low[0], e_low_prev, e_links_low, e_links_high, f_back_low, f_back_high)
  *                 self.find_projection(e_high_prev, e_high[0], e_links_low, e_links_high, f_back_low, f_back_high)             # <<<<<<<<<<<<<<
  * 
  *             if f_back_low[0] > f_low:
  */
-      __pyx_t_2 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->find_projection(__pyx_v_self, __pyx_v_e_high_prev, (__pyx_v_e_high[0]), __pyx_v_e_links_low, __pyx_v_e_links_high, __pyx_v_f_back_low, __pyx_v_f_back_high); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1194; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->find_projection(__pyx_v_self, __pyx_v_e_high_prev, (__pyx_v_e_high[0]), __pyx_v_e_links_low, __pyx_v_e_links_high, __pyx_v_f_back_low, __pyx_v_f_back_high); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1193; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
     }
     __pyx_L11:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1196
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1195
  *                 self.find_projection(e_high_prev, e_high[0], e_links_low, e_links_high, f_back_low, f_back_high)
  * 
  *             if f_back_low[0] > f_low:             # <<<<<<<<<<<<<<
@@ -49253,7 +49305,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     __pyx_t_4 = ((__pyx_v_f_back_low[0]) > __pyx_v_f_low);
     if (__pyx_t_4) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1197
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1196
  * 
  *             if f_back_low[0] > f_low:
  *                 f_back_low[0] = f_low             # <<<<<<<<<<<<<<
@@ -49265,35 +49317,35 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     }
     __pyx_L12:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1199
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1198
  *                 f_back_low[0] = f_low
  * 
  *             if f_back_high[0] < f_high:             # <<<<<<<<<<<<<<
  *                 f_back_high[0] = f_high
  * 
  */
-    __pyx_t_2 = PyInt_FromLong((__pyx_v_f_back_high[0])); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1199; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyInt_FromLong((__pyx_v_f_back_high[0])); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1198; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_6 = PyObject_RichCompare(__pyx_t_2, __pyx_v_f_high, Py_LT); __Pyx_XGOTREF(__pyx_t_6); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1199; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_6 = PyObject_RichCompare(__pyx_t_2, __pyx_v_f_high, Py_LT); __Pyx_XGOTREF(__pyx_t_6); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1198; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_6); if (unlikely(__pyx_t_4 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1199; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_6); if (unlikely(__pyx_t_4 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1198; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
     if (__pyx_t_4) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1200
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1199
  * 
  *             if f_back_high[0] < f_high:
  *                 f_back_high[0] = f_high             # <<<<<<<<<<<<<<
  * 
  *             if f_back_low[0] == f_low_prev and f_back_high[0] == f_high_prev:
  */
-      __pyx_t_1 = __Pyx_PyInt_AsInt(__pyx_v_f_high); if (unlikely((__pyx_t_1 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1200; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_1 = __Pyx_PyInt_AsInt(__pyx_v_f_high); if (unlikely((__pyx_t_1 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1199; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       (__pyx_v_f_back_high[0]) = __pyx_t_1;
       goto __pyx_L13;
     }
     __pyx_L13:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1202
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1201
  *                 f_back_high[0] = f_high
  * 
  *             if f_back_low[0] == f_low_prev and f_back_high[0] == f_high_prev:             # <<<<<<<<<<<<<<
@@ -49309,7 +49361,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     }
     if (__pyx_t_3) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1203
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1202
  * 
  *             if f_back_low[0] == f_low_prev and f_back_high[0] == f_high_prev:
  *                 return 1             # <<<<<<<<<<<<<<
@@ -49322,7 +49374,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     }
     __pyx_L14:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1205
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1204
  *                 return 1
  * 
  *             if allow_low_x == 0 and f_back_low[0] < f_low:             # <<<<<<<<<<<<<<
@@ -49338,7 +49390,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     }
     if (__pyx_t_5) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1207
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1206
  *             if allow_low_x == 0 and f_back_low[0] < f_low:
  *                 # FAIL: f phrase is not tight
  *                 return 0             # <<<<<<<<<<<<<<
@@ -49351,7 +49403,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     }
     __pyx_L15:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1209
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1208
  *                 return 0
  * 
  *             if f_back_high[0] - f_back_low[0] > max_f_len:             # <<<<<<<<<<<<<<
@@ -49361,7 +49413,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     __pyx_t_5 = (((__pyx_v_f_back_high[0]) - (__pyx_v_f_back_low[0])) > __pyx_v_max_f_len);
     if (__pyx_t_5) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1211
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1210
  *             if f_back_high[0] - f_back_low[0] > max_f_len:
  *                 # FAIL: f back projection is too wide
  *                 return 0             # <<<<<<<<<<<<<<
@@ -49374,7 +49426,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     }
     __pyx_L16:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1213
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1212
  *                 return 0
  * 
  *             if allow_high_x == 0 and f_back_high[0] > f_high:             # <<<<<<<<<<<<<<
@@ -49383,11 +49435,11 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
  */
     __pyx_t_5 = (__pyx_v_allow_high_x == 0);
     if (__pyx_t_5) {
-      __pyx_t_6 = PyInt_FromLong((__pyx_v_f_back_high[0])); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1213; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_6 = PyInt_FromLong((__pyx_v_f_back_high[0])); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1212; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_6);
-      __pyx_t_2 = PyObject_RichCompare(__pyx_t_6, __pyx_v_f_high, Py_GT); __Pyx_XGOTREF(__pyx_t_2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1213; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_RichCompare(__pyx_t_6, __pyx_v_f_high, Py_GT); __Pyx_XGOTREF(__pyx_t_2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1212; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
-      __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_3 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1213; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_3 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1212; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
       __pyx_t_4 = __pyx_t_3;
     } else {
@@ -49395,7 +49447,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     }
     if (__pyx_t_4) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1215
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1214
  *             if allow_high_x == 0 and f_back_high[0] > f_high:
  *                 # FAIL: extension on high side not allowed
  *                 return 0             # <<<<<<<<<<<<<<
@@ -49408,7 +49460,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     }
     __pyx_L17:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1217
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1216
  *                 return 0
  * 
  *             if f_low != f_back_low[0]:             # <<<<<<<<<<<<<<
@@ -49418,7 +49470,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     __pyx_t_4 = (__pyx_v_f_low != (__pyx_v_f_back_low[0]));
     if (__pyx_t_4) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1218
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1217
  * 
  *             if f_low != f_back_low[0]:
  *                 if new_low_x == 0:             # <<<<<<<<<<<<<<
@@ -49428,7 +49480,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
       __pyx_t_4 = (__pyx_v_new_low_x == 0);
       if (__pyx_t_4) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1219
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1218
  *             if f_low != f_back_low[0]:
  *                 if new_low_x == 0:
  *                     if new_x >= max_new_x:             # <<<<<<<<<<<<<<
@@ -49438,7 +49490,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
         __pyx_t_4 = (__pyx_v_new_x >= __pyx_v_max_new_x);
         if (__pyx_t_4) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1221
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1220
  *                     if new_x >= max_new_x:
  *                         # FAIL: extension required on low side violates max # of gaps
  *                         return 0             # <<<<<<<<<<<<<<
@@ -49451,7 +49503,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
         }
         /*else*/ {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1223
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1222
  *                         return 0
  *                     else:
  *                         new_x = new_x + 1             # <<<<<<<<<<<<<<
@@ -49460,7 +49512,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
  */
           __pyx_v_new_x = (__pyx_v_new_x + 1);
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1224
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1223
  *                     else:
  *                         new_x = new_x + 1
  *                         new_low_x = 1             # <<<<<<<<<<<<<<
@@ -49474,7 +49526,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
       }
       __pyx_L19:;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1225
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1224
  *                         new_x = new_x + 1
  *                         new_low_x = 1
  *                 if f_low - f_back_low[0] < min_fx_size:             # <<<<<<<<<<<<<<
@@ -49484,7 +49536,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
       __pyx_t_4 = ((__pyx_v_f_low - (__pyx_v_f_back_low[0])) < __pyx_v_min_fx_size);
       if (__pyx_t_4) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1226
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1225
  *                         new_low_x = 1
  *                 if f_low - f_back_low[0] < min_fx_size:
  *                     f_back_low[0] = f_low - min_fx_size             # <<<<<<<<<<<<<<
@@ -49493,7 +49545,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
  */
         (__pyx_v_f_back_low[0]) = (__pyx_v_f_low - __pyx_v_min_fx_size);
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1227
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1226
  *                 if f_low - f_back_low[0] < min_fx_size:
  *                     f_back_low[0] = f_low - min_fx_size
  *                     if f_back_high[0] - f_back_low[0] > max_f_len:             # <<<<<<<<<<<<<<
@@ -49503,7 +49555,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
         __pyx_t_4 = (((__pyx_v_f_back_high[0]) - (__pyx_v_f_back_low[0])) > __pyx_v_max_f_len);
         if (__pyx_t_4) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1229
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1228
  *                     if f_back_high[0] - f_back_low[0] > max_f_len:
  *                         # FAIL: extension required on low side violates max initial length
  *                         return 0             # <<<<<<<<<<<<<<
@@ -49516,7 +49568,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
         }
         __pyx_L22:;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1230
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1229
  *                         # FAIL: extension required on low side violates max initial length
  *                         return 0
  *                     if f_back_low[0] < 0:             # <<<<<<<<<<<<<<
@@ -49526,7 +49578,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
         __pyx_t_4 = ((__pyx_v_f_back_low[0]) < 0);
         if (__pyx_t_4) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1232
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1231
  *                     if f_back_low[0] < 0:
  *                         # FAIL: extension required on low side violates sentence boundary
  *                         return 0             # <<<<<<<<<<<<<<
@@ -49545,22 +49597,22 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     }
     __pyx_L18:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1234
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1233
  *                         return 0
  * 
  *             if f_high != f_back_high[0]:             # <<<<<<<<<<<<<<
  *                 if new_high_x == 0:
  *                     if new_x >= max_new_x:
  */
-    __pyx_t_2 = PyInt_FromLong((__pyx_v_f_back_high[0])); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1234; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyInt_FromLong((__pyx_v_f_back_high[0])); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1233; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_6 = PyObject_RichCompare(__pyx_v_f_high, __pyx_t_2, Py_NE); __Pyx_XGOTREF(__pyx_t_6); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1234; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_6 = PyObject_RichCompare(__pyx_v_f_high, __pyx_t_2, Py_NE); __Pyx_XGOTREF(__pyx_t_6); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1233; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_6); if (unlikely(__pyx_t_4 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1234; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_6); if (unlikely(__pyx_t_4 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1233; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
     if (__pyx_t_4) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1235
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1234
  * 
  *             if f_high != f_back_high[0]:
  *                 if new_high_x == 0:             # <<<<<<<<<<<<<<
@@ -49570,7 +49622,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
       __pyx_t_4 = (__pyx_v_new_high_x == 0);
       if (__pyx_t_4) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1236
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1235
  *             if f_high != f_back_high[0]:
  *                 if new_high_x == 0:
  *                     if new_x >= max_new_x:             # <<<<<<<<<<<<<<
@@ -49580,7 +49632,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
         __pyx_t_4 = (__pyx_v_new_x >= __pyx_v_max_new_x);
         if (__pyx_t_4) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1238
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1237
  *                     if new_x >= max_new_x:
  *                         # FAIL: extension required on high side violates max # of gaps
  *                         return 0             # <<<<<<<<<<<<<<
@@ -49593,7 +49645,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
         }
         /*else*/ {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1240
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1239
  *                         return 0
  *                     else:
  *                         new_x = new_x + 1             # <<<<<<<<<<<<<<
@@ -49602,7 +49654,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
  */
           __pyx_v_new_x = (__pyx_v_new_x + 1);
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1241
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1240
  *                     else:
  *                         new_x = new_x + 1
  *                         new_high_x = 1             # <<<<<<<<<<<<<<
@@ -49616,44 +49668,44 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
       }
       __pyx_L25:;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1242
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1241
  *                         new_x = new_x + 1
  *                         new_high_x = 1
  *                 if f_back_high[0] - f_high < min_fx_size:             # <<<<<<<<<<<<<<
  *                     f_back_high[0] = f_high + min_fx_size
  *                     if f_back_high[0] - f_back_low[0] > max_f_len:
  */
-      __pyx_t_6 = PyInt_FromLong((__pyx_v_f_back_high[0])); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1242; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_6 = PyInt_FromLong((__pyx_v_f_back_high[0])); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1241; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_6);
-      __pyx_t_2 = PyNumber_Subtract(__pyx_t_6, __pyx_v_f_high); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1242; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyNumber_Subtract(__pyx_t_6, __pyx_v_f_high); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1241; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
       __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
-      __pyx_t_6 = PyInt_FromLong(__pyx_v_min_fx_size); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1242; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_6 = PyInt_FromLong(__pyx_v_min_fx_size); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1241; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_6);
-      __pyx_t_7 = PyObject_RichCompare(__pyx_t_2, __pyx_t_6, Py_LT); __Pyx_XGOTREF(__pyx_t_7); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1242; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_7 = PyObject_RichCompare(__pyx_t_2, __pyx_t_6, Py_LT); __Pyx_XGOTREF(__pyx_t_7); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1241; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
       __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
-      __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_7); if (unlikely(__pyx_t_4 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1242; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_7); if (unlikely(__pyx_t_4 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1241; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
       if (__pyx_t_4) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1243
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1242
  *                         new_high_x = 1
  *                 if f_back_high[0] - f_high < min_fx_size:
  *                     f_back_high[0] = f_high + min_fx_size             # <<<<<<<<<<<<<<
  *                     if f_back_high[0] - f_back_low[0] > max_f_len:
  *                         # FAIL: extension required on high side violates max initial length
  */
-        __pyx_t_7 = PyInt_FromLong(__pyx_v_min_fx_size); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1243; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_7 = PyInt_FromLong(__pyx_v_min_fx_size); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1242; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_7);
-        __pyx_t_6 = PyNumber_Add(__pyx_v_f_high, __pyx_t_7); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1243; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_6 = PyNumber_Add(__pyx_v_f_high, __pyx_t_7); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1242; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_6);
         __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
-        __pyx_t_1 = __Pyx_PyInt_AsInt(__pyx_t_6); if (unlikely((__pyx_t_1 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1243; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_1 = __Pyx_PyInt_AsInt(__pyx_t_6); if (unlikely((__pyx_t_1 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1242; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
         (__pyx_v_f_back_high[0]) = __pyx_t_1;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1244
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1243
  *                 if f_back_high[0] - f_high < min_fx_size:
  *                     f_back_high[0] = f_high + min_fx_size
  *                     if f_back_high[0] - f_back_low[0] > max_f_len:             # <<<<<<<<<<<<<<
@@ -49663,7 +49715,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
         __pyx_t_4 = (((__pyx_v_f_back_high[0]) - (__pyx_v_f_back_low[0])) > __pyx_v_max_f_len);
         if (__pyx_t_4) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1246
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1245
  *                     if f_back_high[0] - f_back_low[0] > max_f_len:
  *                         # FAIL: extension required on high side violates max initial length
  *                         return 0             # <<<<<<<<<<<<<<
@@ -49676,7 +49728,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
         }
         __pyx_L28:;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1247
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1246
  *                         # FAIL: extension required on high side violates max initial length
  *                         return 0
  *                     if f_back_high[0] > f_sent_len:             # <<<<<<<<<<<<<<
@@ -49686,7 +49738,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
         __pyx_t_4 = ((__pyx_v_f_back_high[0]) > __pyx_v_f_sent_len);
         if (__pyx_t_4) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1249
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1248
  *                     if f_back_high[0] > f_sent_len:
  *                         # FAIL: extension required on high side violates sentence boundary
  *                         return 0             # <<<<<<<<<<<<<<
@@ -49705,7 +49757,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     }
     __pyx_L24:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1251
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1250
  *                         return 0
  * 
  *             e_low_prev = e_low[0]             # <<<<<<<<<<<<<<
@@ -49714,7 +49766,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
  */
     __pyx_v_e_low_prev = (__pyx_v_e_low[0]);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1252
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1251
  * 
  *             e_low_prev = e_low[0]
  *             e_high_prev = e_high[0]             # <<<<<<<<<<<<<<
@@ -49723,29 +49775,29 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
  */
     __pyx_v_e_high_prev = (__pyx_v_e_high[0]);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1254
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1253
  *             e_high_prev = e_high[0]
  * 
  *             self.find_projection(f_back_low[0], f_low_prev, f_links_low, f_links_high, e_low, e_high)             # <<<<<<<<<<<<<<
  *             self.find_projection(f_high_prev, f_back_high[0], f_links_low, f_links_high, e_low, e_high)
  *             if e_low[0] == e_low_prev and e_high[0] == e_high_prev:
  */
-    __pyx_t_6 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->find_projection(__pyx_v_self, (__pyx_v_f_back_low[0]), __pyx_v_f_low_prev, __pyx_v_f_links_low, __pyx_v_f_links_high, __pyx_v_e_low, __pyx_v_e_high); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1254; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_6 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->find_projection(__pyx_v_self, (__pyx_v_f_back_low[0]), __pyx_v_f_low_prev, __pyx_v_f_links_low, __pyx_v_f_links_high, __pyx_v_e_low, __pyx_v_e_high); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1253; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_6);
     __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1255
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1254
  * 
  *             self.find_projection(f_back_low[0], f_low_prev, f_links_low, f_links_high, e_low, e_high)
  *             self.find_projection(f_high_prev, f_back_high[0], f_links_low, f_links_high, e_low, e_high)             # <<<<<<<<<<<<<<
  *             if e_low[0] == e_low_prev and e_high[0] == e_high_prev:
  *                 return 1
  */
-    __pyx_t_6 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->find_projection(__pyx_v_self, __pyx_v_f_high_prev, (__pyx_v_f_back_high[0]), __pyx_v_f_links_low, __pyx_v_f_links_high, __pyx_v_e_low, __pyx_v_e_high); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1255; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_6 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->find_projection(__pyx_v_self, __pyx_v_f_high_prev, (__pyx_v_f_back_high[0]), __pyx_v_f_links_low, __pyx_v_f_links_high, __pyx_v_e_low, __pyx_v_e_high); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1254; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_6);
     __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1256
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1255
  *             self.find_projection(f_back_low[0], f_low_prev, f_links_low, f_links_high, e_low, e_high)
  *             self.find_projection(f_high_prev, f_back_high[0], f_links_low, f_links_high, e_low, e_high)
  *             if e_low[0] == e_low_prev and e_high[0] == e_high_prev:             # <<<<<<<<<<<<<<
@@ -49761,7 +49813,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     }
     if (__pyx_t_3) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1257
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1256
  *             self.find_projection(f_high_prev, f_back_high[0], f_links_low, f_links_high, e_low, e_high)
  *             if e_low[0] == e_low_prev and e_high[0] == e_high_prev:
  *                 return 1             # <<<<<<<<<<<<<<
@@ -49774,7 +49826,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     }
     __pyx_L30:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1258
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1257
  *             if e_low[0] == e_low_prev and e_high[0] == e_high_prev:
  *                 return 1
  *             if allow_arbitrary_x == 0:             # <<<<<<<<<<<<<<
@@ -49784,7 +49836,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     __pyx_t_3 = (__pyx_v_allow_arbitrary_x == 0);
     if (__pyx_t_3) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1260
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1259
  *             if allow_arbitrary_x == 0:
  *                 # FAIL: arbitrary expansion not permitted
  *                 return 0             # <<<<<<<<<<<<<<
@@ -49797,7 +49849,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     }
     __pyx_L31:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1261
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1260
  *                 # FAIL: arbitrary expansion not permitted
  *                 return 0
  *             if e_high[0] - e_low[0] > max_e_len:             # <<<<<<<<<<<<<<
@@ -49807,7 +49859,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     __pyx_t_3 = (((__pyx_v_e_high[0]) - (__pyx_v_e_low[0])) > __pyx_v_max_e_len);
     if (__pyx_t_3) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1263
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1262
  *             if e_high[0] - e_low[0] > max_e_len:
  *                 # FAIL: re-projection violates sentence max phrase length
  *                 return 0             # <<<<<<<<<<<<<<
@@ -49820,7 +49872,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
     }
     __pyx_L32:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1264
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1263
  *                 # FAIL: re-projection violates sentence max phrase length
  *                 return 0
  *             f_low_prev = f_back_low[0]             # <<<<<<<<<<<<<<
@@ -49829,7 +49881,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
  */
     __pyx_v_f_low_prev = (__pyx_v_f_back_low[0]);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1265
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1264
  *                 return 0
  *             f_low_prev = f_back_low[0]
  *             f_high_prev = f_back_high[0]             # <<<<<<<<<<<<<<
@@ -49852,7 +49904,7 @@ static int __pyx_f_3_sa_23HieroCachingRuleFactory_find_fixpoint(struct __pyx_obj
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1268
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1267
  * 
  * 
  *     cdef find_projection(self, int in_low, int in_high, int* in_links_low, int* in_links_high,             # <<<<<<<<<<<<<<
@@ -49870,7 +49922,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_find_projection(CYTHON_U
   int __pyx_t_4;
   __Pyx_RefNannySetupContext("find_projection", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1271
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1270
  *                         int* out_low, int* out_high):
  *         cdef int i
  *         for i from in_low <= i < in_high:             # <<<<<<<<<<<<<<
@@ -49880,7 +49932,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_find_projection(CYTHON_U
   __pyx_t_1 = __pyx_v_in_high;
   for (__pyx_v_i = __pyx_v_in_low; __pyx_v_i < __pyx_t_1; __pyx_v_i++) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1272
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1271
  *         cdef int i
  *         for i from in_low <= i < in_high:
  *             if in_links_low[i] != -1:             # <<<<<<<<<<<<<<
@@ -49890,7 +49942,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_find_projection(CYTHON_U
     __pyx_t_2 = ((__pyx_v_in_links_low[__pyx_v_i]) != -1);
     if (__pyx_t_2) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1273
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1272
  *         for i from in_low <= i < in_high:
  *             if in_links_low[i] != -1:
  *                 if out_low[0] == -1 or in_links_low[i] < out_low[0]:             # <<<<<<<<<<<<<<
@@ -49906,7 +49958,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_find_projection(CYTHON_U
       }
       if (__pyx_t_4) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1274
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1273
  *             if in_links_low[i] != -1:
  *                 if out_low[0] == -1 or in_links_low[i] < out_low[0]:
  *                     out_low[0] = in_links_low[i]             # <<<<<<<<<<<<<<
@@ -49918,7 +49970,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_find_projection(CYTHON_U
       }
       __pyx_L6:;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1275
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1274
  *                 if out_low[0] == -1 or in_links_low[i] < out_low[0]:
  *                     out_low[0] = in_links_low[i]
  *                 if out_high[0] == -1 or in_links_high[i] > out_high[0]:             # <<<<<<<<<<<<<<
@@ -49934,7 +49986,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_find_projection(CYTHON_U
       }
       if (__pyx_t_3) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1276
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1275
  *                     out_low[0] = in_links_low[i]
  *                 if out_high[0] == -1 or in_links_high[i] > out_high[0]:
  *                     out_high[0] = in_links_high[i]             # <<<<<<<<<<<<<<
@@ -49956,7 +50008,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_find_projection(CYTHON_U
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1279
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1278
  * 
  * 
  *     cdef int* int_arr_extend(self, int* arr, int* arr_len, int* data, int data_len):             # <<<<<<<<<<<<<<
@@ -49970,7 +50022,7 @@ static int *__pyx_f_3_sa_23HieroCachingRuleFactory_int_arr_extend(CYTHON_UNUSED
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("int_arr_extend", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1281
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1280
  *     cdef int* int_arr_extend(self, int* arr, int* arr_len, int* data, int data_len):
  *         cdef int new_len
  *         new_len = arr_len[0] + data_len             # <<<<<<<<<<<<<<
@@ -49979,7 +50031,7 @@ static int *__pyx_f_3_sa_23HieroCachingRuleFactory_int_arr_extend(CYTHON_UNUSED
  */
   __pyx_v_new_len = ((__pyx_v_arr_len[0]) + __pyx_v_data_len);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1282
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1281
  *         cdef int new_len
  *         new_len = arr_len[0] + data_len
  *         arr = <int*> realloc(arr, new_len*sizeof(int))             # <<<<<<<<<<<<<<
@@ -49988,7 +50040,7 @@ static int *__pyx_f_3_sa_23HieroCachingRuleFactory_int_arr_extend(CYTHON_UNUSED
  */
   __pyx_v_arr = ((int *)realloc(__pyx_v_arr, (__pyx_v_new_len * (sizeof(int)))));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1283
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1282
  *         new_len = arr_len[0] + data_len
  *         arr = <int*> realloc(arr, new_len*sizeof(int))
  *         memcpy(arr+arr_len[0], data, data_len*sizeof(int))             # <<<<<<<<<<<<<<
@@ -49997,7 +50049,7 @@ static int *__pyx_f_3_sa_23HieroCachingRuleFactory_int_arr_extend(CYTHON_UNUSED
  */
   memcpy((__pyx_v_arr + (__pyx_v_arr_len[0])), __pyx_v_data, (__pyx_v_data_len * (sizeof(int))));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1284
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1283
  *         arr = <int*> realloc(arr, new_len*sizeof(int))
  *         memcpy(arr+arr_len[0], data, data_len*sizeof(int))
  *         arr_len[0] = new_len             # <<<<<<<<<<<<<<
@@ -50006,7 +50058,7 @@ static int *__pyx_f_3_sa_23HieroCachingRuleFactory_int_arr_extend(CYTHON_UNUSED
  */
   (__pyx_v_arr_len[0]) = __pyx_v_new_len;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1285
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1284
  *         memcpy(arr+arr_len[0], data, data_len*sizeof(int))
  *         arr_len[0] = new_len
  *         return arr             # <<<<<<<<<<<<<<
@@ -50022,7 +50074,7 @@ static int *__pyx_f_3_sa_23HieroCachingRuleFactory_int_arr_extend(CYTHON_UNUSED
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1288
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1287
  * 
  * 
  *     cdef extract_phrases(self, int e_low, int e_high, int* e_gap_low, int* e_gap_high, int* e_links_low, int num_gaps,             # <<<<<<<<<<<<<<
@@ -50068,19 +50120,19 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("extract_phrases", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1296
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1295
  *         cdef result
  * 
  *         result = []             # <<<<<<<<<<<<<<
  *         len1 = 0
  *         e_gaps1 = <int*> malloc(0)
  */
-  __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1296; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1295; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __pyx_v_result = ((PyObject *)__pyx_t_1);
   __pyx_t_1 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1297
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1296
  * 
  *         result = []
  *         len1 = 0             # <<<<<<<<<<<<<<
@@ -50089,7 +50141,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
   __pyx_v_len1 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1298
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1297
  *         result = []
  *         len1 = 0
  *         e_gaps1 = <int*> malloc(0)             # <<<<<<<<<<<<<<
@@ -50098,19 +50150,19 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
   __pyx_v_e_gaps1 = ((int *)malloc(0));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1299
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1298
  *         len1 = 0
  *         e_gaps1 = <int*> malloc(0)
  *         ephr_arr = IntList()             # <<<<<<<<<<<<<<
  * 
  *         e_gap_order = <int*> malloc(num_gaps*sizeof(int))
  */
-  __pyx_t_1 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_IntList)), ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1299; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_IntList)), ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1298; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __pyx_v_ephr_arr = ((struct __pyx_obj_3_sa_IntList *)__pyx_t_1);
   __pyx_t_1 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1301
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1300
  *         ephr_arr = IntList()
  * 
  *         e_gap_order = <int*> malloc(num_gaps*sizeof(int))             # <<<<<<<<<<<<<<
@@ -50119,7 +50171,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
   __pyx_v_e_gap_order = ((int *)malloc((__pyx_v_num_gaps * (sizeof(int)))));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1302
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1301
  * 
  *         e_gap_order = <int*> malloc(num_gaps*sizeof(int))
  *         if num_gaps > 0:             # <<<<<<<<<<<<<<
@@ -50129,7 +50181,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
   __pyx_t_2 = (__pyx_v_num_gaps > 0);
   if (__pyx_t_2) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1303
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1302
  *         e_gap_order = <int*> malloc(num_gaps*sizeof(int))
  *         if num_gaps > 0:
  *             e_gap_order[0] = 0             # <<<<<<<<<<<<<<
@@ -50138,7 +50190,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
     (__pyx_v_e_gap_order[0]) = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1304
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1303
  *         if num_gaps > 0:
  *             e_gap_order[0] = 0
  *             for i from 1 <= i < num_gaps:             # <<<<<<<<<<<<<<
@@ -50148,7 +50200,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
     __pyx_t_3 = __pyx_v_num_gaps;
     for (__pyx_v_i = 1; __pyx_v_i < __pyx_t_3; __pyx_v_i++) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1305
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1304
  *             e_gap_order[0] = 0
  *             for i from 1 <= i < num_gaps:
  *                 for j from 0 <= j < i:             # <<<<<<<<<<<<<<
@@ -50158,7 +50210,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
       __pyx_t_4 = __pyx_v_i;
       for (__pyx_v_j = 0; __pyx_v_j < __pyx_t_4; __pyx_v_j++) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1306
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1305
  *             for i from 1 <= i < num_gaps:
  *                 for j from 0 <= j < i:
  *                     if e_gap_low[i] < e_gap_low[j]:             # <<<<<<<<<<<<<<
@@ -50168,7 +50220,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
         __pyx_t_2 = ((__pyx_v_e_gap_low[__pyx_v_i]) < (__pyx_v_e_gap_low[__pyx_v_j]));
         if (__pyx_t_2) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1307
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1306
  *                 for j from 0 <= j < i:
  *                     if e_gap_low[i] < e_gap_low[j]:
  *                         for k from j <= k < i:             # <<<<<<<<<<<<<<
@@ -50178,7 +50230,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
           __pyx_t_5 = __pyx_v_i;
           for (__pyx_v_k = __pyx_v_j; __pyx_v_k < __pyx_t_5; __pyx_v_k++) {
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1308
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1307
  *                     if e_gap_low[i] < e_gap_low[j]:
  *                         for k from j <= k < i:
  *                             e_gap_order[k+1] = e_gap_order[k]             # <<<<<<<<<<<<<<
@@ -50188,7 +50240,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
             (__pyx_v_e_gap_order[(__pyx_v_k + 1)]) = (__pyx_v_e_gap_order[__pyx_v_k]);
           }
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1309
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1308
  *                         for k from j <= k < i:
  *                             e_gap_order[k+1] = e_gap_order[k]
  *                         e_gap_order[j] = i             # <<<<<<<<<<<<<<
@@ -50197,7 +50249,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
           (__pyx_v_e_gap_order[__pyx_v_j]) = __pyx_v_i;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1310
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1309
  *                             e_gap_order[k+1] = e_gap_order[k]
  *                         e_gap_order[j] = i
  *                         break             # <<<<<<<<<<<<<<
@@ -50211,7 +50263,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
       }
       /*else*/ {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1312
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1311
  *                         break
  *                 else:
  *                     e_gap_order[i] = i             # <<<<<<<<<<<<<<
@@ -50226,7 +50278,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
   }
   __pyx_L3:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1314
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1313
  *                     e_gap_order[i] = i
  * 
  *         e_x_low = e_low             # <<<<<<<<<<<<<<
@@ -50235,7 +50287,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
   __pyx_v_e_x_low = __pyx_v_e_low;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1315
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1314
  * 
  *         e_x_low = e_low
  *         e_x_high = e_high             # <<<<<<<<<<<<<<
@@ -50244,7 +50296,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
   __pyx_v_e_x_high = __pyx_v_e_high;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1316
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1315
  *         e_x_low = e_low
  *         e_x_high = e_high
  *         if self.tight_phrases == 0:             # <<<<<<<<<<<<<<
@@ -50254,7 +50306,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
   __pyx_t_2 = (__pyx_v_self->tight_phrases == 0);
   if (__pyx_t_2) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1317
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1316
  *         e_x_high = e_high
  *         if self.tight_phrases == 0:
  *             while e_x_low > 0 and e_high - e_x_low < self.train_max_initial_size and e_links_low[e_x_low-1] == -1:             # <<<<<<<<<<<<<<
@@ -50277,7 +50329,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
       }
       if (!__pyx_t_6) break;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1318
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1317
  *         if self.tight_phrases == 0:
  *             while e_x_low > 0 and e_high - e_x_low < self.train_max_initial_size and e_links_low[e_x_low-1] == -1:
  *                 e_x_low = e_x_low - 1             # <<<<<<<<<<<<<<
@@ -50287,7 +50339,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
       __pyx_v_e_x_low = (__pyx_v_e_x_low - 1);
     }
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1319
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1318
  *             while e_x_low > 0 and e_high - e_x_low < self.train_max_initial_size and e_links_low[e_x_low-1] == -1:
  *                 e_x_low = e_x_low - 1
  *             while e_x_high < e_sent_len and e_x_high - e_low < self.train_max_initial_size and e_links_low[e_x_high] == -1:             # <<<<<<<<<<<<<<
@@ -50310,7 +50362,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
       }
       if (!__pyx_t_2) break;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1320
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1319
  *                 e_x_low = e_x_low - 1
  *             while e_x_high < e_sent_len and e_x_high - e_low < self.train_max_initial_size and e_links_low[e_x_high] == -1:
  *                 e_x_high = e_x_high + 1             # <<<<<<<<<<<<<<
@@ -50323,7 +50375,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
   }
   __pyx_L11:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1322
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1321
  *                 e_x_high = e_x_high + 1
  * 
  *         for i from e_x_low <= i <= e_low:             # <<<<<<<<<<<<<<
@@ -50333,7 +50385,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
   __pyx_t_3 = __pyx_v_e_low;
   for (__pyx_v_i = __pyx_v_e_x_low; __pyx_v_i <= __pyx_t_3; __pyx_v_i++) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1323
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1322
  * 
  *         for i from e_x_low <= i <= e_low:
  *             e_gaps1 = self.int_arr_extend(e_gaps1, &len1, &i, 1)             # <<<<<<<<<<<<<<
@@ -50343,7 +50395,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
     __pyx_v_e_gaps1 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->int_arr_extend(__pyx_v_self, __pyx_v_e_gaps1, (&__pyx_v_len1), (&__pyx_v_i), 1);
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1325
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1324
  *             e_gaps1 = self.int_arr_extend(e_gaps1, &len1, &i, 1)
  * 
  *         for i from 0 <= i < num_gaps:             # <<<<<<<<<<<<<<
@@ -50353,7 +50405,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
   __pyx_t_3 = __pyx_v_num_gaps;
   for (__pyx_v_i = 0; __pyx_v_i < __pyx_t_3; __pyx_v_i++) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1326
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1325
  * 
  *         for i from 0 <= i < num_gaps:
  *             e_gaps2 = <int*> malloc(0)             # <<<<<<<<<<<<<<
@@ -50362,7 +50414,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
     __pyx_v_e_gaps2 = ((int *)malloc(0));
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1327
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1326
  *         for i from 0 <= i < num_gaps:
  *             e_gaps2 = <int*> malloc(0)
  *             len2 = 0             # <<<<<<<<<<<<<<
@@ -50371,7 +50423,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
     __pyx_v_len2 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1329
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1328
  *             len2 = 0
  * 
  *             j = e_gap_order[i]             # <<<<<<<<<<<<<<
@@ -50380,7 +50432,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
     __pyx_v_j = (__pyx_v_e_gap_order[__pyx_v_i]);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1330
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1329
  * 
  *             j = e_gap_order[i]
  *             e_x_gap_low = e_gap_low[j]             # <<<<<<<<<<<<<<
@@ -50389,7 +50441,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
     __pyx_v_e_x_gap_low = (__pyx_v_e_gap_low[__pyx_v_j]);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1331
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1330
  *             j = e_gap_order[i]
  *             e_x_gap_low = e_gap_low[j]
  *             e_x_gap_high = e_gap_high[j]             # <<<<<<<<<<<<<<
@@ -50398,7 +50450,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
     __pyx_v_e_x_gap_high = (__pyx_v_e_gap_high[__pyx_v_j]);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1332
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1331
  *             e_x_gap_low = e_gap_low[j]
  *             e_x_gap_high = e_gap_high[j]
  *             if self.tight_phrases == 0:             # <<<<<<<<<<<<<<
@@ -50408,7 +50460,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
     __pyx_t_2 = (__pyx_v_self->tight_phrases == 0);
     if (__pyx_t_2) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1333
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1332
  *             e_x_gap_high = e_gap_high[j]
  *             if self.tight_phrases == 0:
  *                 while e_x_gap_low > e_x_low and e_links_low[e_x_gap_low-1] == -1:             # <<<<<<<<<<<<<<
@@ -50425,7 +50477,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
         }
         if (!__pyx_t_7) break;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1334
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1333
  *             if self.tight_phrases == 0:
  *                 while e_x_gap_low > e_x_low and e_links_low[e_x_gap_low-1] == -1:
  *                     e_x_gap_low = e_x_gap_low - 1             # <<<<<<<<<<<<<<
@@ -50435,7 +50487,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
         __pyx_v_e_x_gap_low = (__pyx_v_e_x_gap_low - 1);
       }
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1335
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1334
  *                 while e_x_gap_low > e_x_low and e_links_low[e_x_gap_low-1] == -1:
  *                     e_x_gap_low = e_x_gap_low - 1
  *                 while e_x_gap_high < e_x_high and e_links_low[e_x_gap_high] == -1:             # <<<<<<<<<<<<<<
@@ -50452,7 +50504,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
         }
         if (!__pyx_t_6) break;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1336
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1335
  *                     e_x_gap_low = e_x_gap_low - 1
  *                 while e_x_gap_high < e_x_high and e_links_low[e_x_gap_high] == -1:
  *                     e_x_gap_high = e_x_gap_high + 1             # <<<<<<<<<<<<<<
@@ -50465,7 +50517,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
     }
     __pyx_L20:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1338
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1337
  *                     e_x_gap_high = e_x_gap_high + 1
  * 
  *             k = 0             # <<<<<<<<<<<<<<
@@ -50474,7 +50526,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
     __pyx_v_k = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1339
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1338
  * 
  *             k = 0
  *             step = 1+(i*2)             # <<<<<<<<<<<<<<
@@ -50483,7 +50535,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
     __pyx_v_step = (1 + (__pyx_v_i * 2));
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1340
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1339
  *             k = 0
  *             step = 1+(i*2)
  *             while k < len1:             # <<<<<<<<<<<<<<
@@ -50494,7 +50546,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
       __pyx_t_6 = (__pyx_v_k < __pyx_v_len1);
       if (!__pyx_t_6) break;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1341
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1340
  *             step = 1+(i*2)
  *             while k < len1:
  *                 for m from e_x_gap_low <= m <= e_gap_low[j]:             # <<<<<<<<<<<<<<
@@ -50504,7 +50556,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
       __pyx_t_4 = (__pyx_v_e_gap_low[__pyx_v_j]);
       for (__pyx_v_m = __pyx_v_e_x_gap_low; __pyx_v_m <= __pyx_t_4; __pyx_v_m++) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1342
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1341
  *             while k < len1:
  *                 for m from e_x_gap_low <= m <= e_gap_low[j]:
  *                     if m >= e_gaps1[k+step-1]:             # <<<<<<<<<<<<<<
@@ -50514,7 +50566,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
         __pyx_t_6 = (__pyx_v_m >= (__pyx_v_e_gaps1[((__pyx_v_k + __pyx_v_step) - 1)]));
         if (__pyx_t_6) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1343
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1342
  *                 for m from e_x_gap_low <= m <= e_gap_low[j]:
  *                     if m >= e_gaps1[k+step-1]:
  *                         for n from e_gap_high[j] <= n <= e_x_gap_high:             # <<<<<<<<<<<<<<
@@ -50524,7 +50576,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
           __pyx_t_5 = __pyx_v_e_x_gap_high;
           for (__pyx_v_n = (__pyx_v_e_gap_high[__pyx_v_j]); __pyx_v_n <= __pyx_t_5; __pyx_v_n++) {
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1344
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1343
  *                     if m >= e_gaps1[k+step-1]:
  *                         for n from e_gap_high[j] <= n <= e_x_gap_high:
  *                             if n-m >= 1:    # extractor.py doesn't restrict target-side gap length             # <<<<<<<<<<<<<<
@@ -50534,7 +50586,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
             __pyx_t_6 = ((__pyx_v_n - __pyx_v_m) >= 1);
             if (__pyx_t_6) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1345
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1344
  *                         for n from e_gap_high[j] <= n <= e_x_gap_high:
  *                             if n-m >= 1:    # extractor.py doesn't restrict target-side gap length
  *                                 e_gaps2 = self.int_arr_extend(e_gaps2, &len2, e_gaps1+k, step)             # <<<<<<<<<<<<<<
@@ -50543,7 +50595,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
               __pyx_v_e_gaps2 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->int_arr_extend(__pyx_v_self, __pyx_v_e_gaps2, (&__pyx_v_len2), (__pyx_v_e_gaps1 + __pyx_v_k), __pyx_v_step);
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1346
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1345
  *                             if n-m >= 1:    # extractor.py doesn't restrict target-side gap length
  *                                 e_gaps2 = self.int_arr_extend(e_gaps2, &len2, e_gaps1+k, step)
  *                                 e_gaps2 = self.int_arr_extend(e_gaps2, &len2, &m, 1)             # <<<<<<<<<<<<<<
@@ -50552,7 +50604,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
               __pyx_v_e_gaps2 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->int_arr_extend(__pyx_v_self, __pyx_v_e_gaps2, (&__pyx_v_len2), (&__pyx_v_m), 1);
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1347
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1346
  *                                 e_gaps2 = self.int_arr_extend(e_gaps2, &len2, e_gaps1+k, step)
  *                                 e_gaps2 = self.int_arr_extend(e_gaps2, &len2, &m, 1)
  *                                 e_gaps2 = self.int_arr_extend(e_gaps2, &len2, &n, 1)             # <<<<<<<<<<<<<<
@@ -50569,7 +50621,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
         __pyx_L29:;
       }
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1348
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1347
  *                                 e_gaps2 = self.int_arr_extend(e_gaps2, &len2, &m, 1)
  *                                 e_gaps2 = self.int_arr_extend(e_gaps2, &len2, &n, 1)
  *                 k = k + step             # <<<<<<<<<<<<<<
@@ -50579,7 +50631,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
       __pyx_v_k = (__pyx_v_k + __pyx_v_step);
     }
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1349
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1348
  *                                 e_gaps2 = self.int_arr_extend(e_gaps2, &len2, &n, 1)
  *                 k = k + step
  *             free(e_gaps1)             # <<<<<<<<<<<<<<
@@ -50588,7 +50640,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
     free(__pyx_v_e_gaps1);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1350
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1349
  *                 k = k + step
  *             free(e_gaps1)
  *             e_gaps1 = e_gaps2             # <<<<<<<<<<<<<<
@@ -50597,7 +50649,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
     __pyx_v_e_gaps1 = __pyx_v_e_gaps2;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1351
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1350
  *             free(e_gaps1)
  *             e_gaps1 = e_gaps2
  *             len1 = len2             # <<<<<<<<<<<<<<
@@ -50607,7 +50659,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
     __pyx_v_len1 = __pyx_v_len2;
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1353
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1352
  *             len1 = len2
  * 
  *         step = 1+(num_gaps*2)             # <<<<<<<<<<<<<<
@@ -50616,7 +50668,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
   __pyx_v_step = (1 + (__pyx_v_num_gaps * 2));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1354
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1353
  * 
  *         step = 1+(num_gaps*2)
  *         e_gaps2 = <int*> malloc(0)             # <<<<<<<<<<<<<<
@@ -50625,7 +50677,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
   __pyx_v_e_gaps2 = ((int *)malloc(0));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1355
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1354
  *         step = 1+(num_gaps*2)
  *         e_gaps2 = <int*> malloc(0)
  *         len2 = 0             # <<<<<<<<<<<<<<
@@ -50634,7 +50686,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
   __pyx_v_len2 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1356
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1355
  *         e_gaps2 = <int*> malloc(0)
  *         len2 = 0
  *         for i from e_high <= i <= e_x_high:             # <<<<<<<<<<<<<<
@@ -50644,7 +50696,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
   __pyx_t_3 = __pyx_v_e_x_high;
   for (__pyx_v_i = __pyx_v_e_high; __pyx_v_i <= __pyx_t_3; __pyx_v_i++) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1357
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1356
  *         len2 = 0
  *         for i from e_high <= i <= e_x_high:
  *             j = 0             # <<<<<<<<<<<<<<
@@ -50653,7 +50705,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
     __pyx_v_j = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1358
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1357
  *         for i from e_high <= i <= e_x_high:
  *             j = 0
  *             while j < len1:             # <<<<<<<<<<<<<<
@@ -50664,7 +50716,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
       __pyx_t_6 = (__pyx_v_j < __pyx_v_len1);
       if (!__pyx_t_6) break;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1359
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1358
  *             j = 0
  *             while j < len1:
  *                 if i - e_gaps1[j] <= self.train_max_initial_size and i >= e_gaps1[j+step-1]:             # <<<<<<<<<<<<<<
@@ -50680,7 +50732,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
       }
       if (__pyx_t_2) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1360
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1359
  *             while j < len1:
  *                 if i - e_gaps1[j] <= self.train_max_initial_size and i >= e_gaps1[j+step-1]:
  *                     e_gaps2 = self.int_arr_extend(e_gaps2, &len2, e_gaps1+j, step)             # <<<<<<<<<<<<<<
@@ -50689,7 +50741,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
         __pyx_v_e_gaps2 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->int_arr_extend(__pyx_v_self, __pyx_v_e_gaps2, (&__pyx_v_len2), (__pyx_v_e_gaps1 + __pyx_v_j), __pyx_v_step);
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1361
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1360
  *                 if i - e_gaps1[j] <= self.train_max_initial_size and i >= e_gaps1[j+step-1]:
  *                     e_gaps2 = self.int_arr_extend(e_gaps2, &len2, e_gaps1+j, step)
  *                     e_gaps2 = self.int_arr_extend(e_gaps2, &len2, &i, 1)             # <<<<<<<<<<<<<<
@@ -50701,7 +50753,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
       }
       __pyx_L37:;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1362
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1361
  *                     e_gaps2 = self.int_arr_extend(e_gaps2, &len2, e_gaps1+j, step)
  *                     e_gaps2 = self.int_arr_extend(e_gaps2, &len2, &i, 1)
  *                 j = j + step             # <<<<<<<<<<<<<<
@@ -50712,7 +50764,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
     }
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1363
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1362
  *                     e_gaps2 = self.int_arr_extend(e_gaps2, &len2, &i, 1)
  *                 j = j + step
  *         free(e_gaps1)             # <<<<<<<<<<<<<<
@@ -50721,7 +50773,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
   free(__pyx_v_e_gaps1);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1364
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1363
  *                 j = j + step
  *         free(e_gaps1)
  *         e_gaps1 = e_gaps2             # <<<<<<<<<<<<<<
@@ -50730,7 +50782,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
   __pyx_v_e_gaps1 = __pyx_v_e_gaps2;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1365
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1364
  *         free(e_gaps1)
  *         e_gaps1 = e_gaps2
  *         len1 = len2             # <<<<<<<<<<<<<<
@@ -50739,7 +50791,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
   __pyx_v_len1 = __pyx_v_len2;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1367
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1366
  *         len1 = len2
  * 
  *         step = (num_gaps+1)*2             # <<<<<<<<<<<<<<
@@ -50748,7 +50800,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
   __pyx_v_step = ((__pyx_v_num_gaps + 1) * 2);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1368
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1367
  * 
  *         step = (num_gaps+1)*2
  *         i = 0             # <<<<<<<<<<<<<<
@@ -50757,7 +50809,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
   __pyx_v_i = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1370
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1369
  *         i = 0
  * 
  *         while i < len1:             # <<<<<<<<<<<<<<
@@ -50768,7 +50820,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
     __pyx_t_2 = (__pyx_v_i < __pyx_v_len1);
     if (!__pyx_t_2) break;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1371
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1370
  * 
  *         while i < len1:
  *             ephr_arr._clear()             # <<<<<<<<<<<<<<
@@ -50777,7 +50829,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
     ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_ephr_arr->__pyx_vtab)->_clear(__pyx_v_ephr_arr);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1372
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1371
  *         while i < len1:
  *             ephr_arr._clear()
  *             num_chunks = 0             # <<<<<<<<<<<<<<
@@ -50786,20 +50838,20 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
     __pyx_v_num_chunks = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1373
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1372
  *             ephr_arr._clear()
  *             num_chunks = 0
  *             indexes = []             # <<<<<<<<<<<<<<
  *             for j from 0 <= j < num_gaps+1:
  *                 if e_gaps1[i+2*j] < e_gaps1[i+(2*j)+1]:
  */
-    __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1373; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1372; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
     __Pyx_XDECREF(((PyObject *)__pyx_v_indexes));
     __pyx_v_indexes = __pyx_t_1;
     __pyx_t_1 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1374
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1373
  *             num_chunks = 0
  *             indexes = []
  *             for j from 0 <= j < num_gaps+1:             # <<<<<<<<<<<<<<
@@ -50809,7 +50861,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
     __pyx_t_9 = (__pyx_v_num_gaps + 1);
     for (__pyx_v_j = 0; __pyx_v_j < __pyx_t_9; __pyx_v_j++) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1375
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1374
  *             indexes = []
  *             for j from 0 <= j < num_gaps+1:
  *                 if e_gaps1[i+2*j] < e_gaps1[i+(2*j)+1]:             # <<<<<<<<<<<<<<
@@ -50819,7 +50871,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
       __pyx_t_2 = ((__pyx_v_e_gaps1[(__pyx_v_i + (2 * __pyx_v_j))]) < (__pyx_v_e_gaps1[((__pyx_v_i + (2 * __pyx_v_j)) + 1)]));
       if (__pyx_t_2) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1376
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1375
  *             for j from 0 <= j < num_gaps+1:
  *                 if e_gaps1[i+2*j] < e_gaps1[i+(2*j)+1]:
  *                     num_chunks = num_chunks + 1             # <<<<<<<<<<<<<<
@@ -50831,7 +50883,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
       }
       __pyx_L42:;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1377
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1376
  *                 if e_gaps1[i+2*j] < e_gaps1[i+(2*j)+1]:
  *                     num_chunks = num_chunks + 1
  *                 for k from e_gaps1[i+2*j] <= k < e_gaps1[i+(2*j)+1]:             # <<<<<<<<<<<<<<
@@ -50841,19 +50893,19 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
       __pyx_t_3 = (__pyx_v_e_gaps1[((__pyx_v_i + (2 * __pyx_v_j)) + 1)]);
       for (__pyx_v_k = (__pyx_v_e_gaps1[(__pyx_v_i + (2 * __pyx_v_j))]); __pyx_v_k < __pyx_t_3; __pyx_v_k++) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1378
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1377
  *                     num_chunks = num_chunks + 1
  *                 for k from e_gaps1[i+2*j] <= k < e_gaps1[i+(2*j)+1]:
  *                     indexes.append(k)             # <<<<<<<<<<<<<<
  *                     ephr_arr._append(self.eid2symid[self.eda.data.arr[e_sent_start+k]])
  *                 if j < num_gaps:
  */
-        __pyx_t_1 = PyInt_FromLong(__pyx_v_k); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1378; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_1 = PyInt_FromLong(__pyx_v_k); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1377; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_1);
-        __pyx_t_10 = PyList_Append(__pyx_v_indexes, __pyx_t_1); if (unlikely(__pyx_t_10 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1378; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_10 = PyList_Append(__pyx_v_indexes, __pyx_t_1); if (unlikely(__pyx_t_10 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1377; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1379
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1378
  *                 for k from e_gaps1[i+2*j] <= k < e_gaps1[i+(2*j)+1]:
  *                     indexes.append(k)
  *                     ephr_arr._append(self.eid2symid[self.eda.data.arr[e_sent_start+k]])             # <<<<<<<<<<<<<<
@@ -50861,14 +50913,14 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  *                     indexes.append(sym_setindex(self.category, e_gap_order[j]+1))
  */
         __pyx_t_4 = (__pyx_v_self->eda->data->arr[(__pyx_v_e_sent_start + __pyx_v_k)]);
-        __pyx_t_1 = __Pyx_GetItemInt(((PyObject *)__pyx_v_self->eid2symid), __pyx_t_4, sizeof(int), PyInt_FromLong); if (!__pyx_t_1) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1379; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_1 = __Pyx_GetItemInt(((PyObject *)__pyx_v_self->eid2symid), __pyx_t_4, sizeof(int), PyInt_FromLong); if (!__pyx_t_1) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1378; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_1);
-        __pyx_t_4 = __Pyx_PyInt_AsInt(__pyx_t_1); if (unlikely((__pyx_t_4 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1379; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_4 = __Pyx_PyInt_AsInt(__pyx_t_1); if (unlikely((__pyx_t_4 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1378; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
         ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_ephr_arr->__pyx_vtab)->_append(__pyx_v_ephr_arr, __pyx_t_4);
       }
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1380
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1379
  *                     indexes.append(k)
  *                     ephr_arr._append(self.eid2symid[self.eda.data.arr[e_sent_start+k]])
  *                 if j < num_gaps:             # <<<<<<<<<<<<<<
@@ -50878,19 +50930,19 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
       __pyx_t_2 = (__pyx_v_j < __pyx_v_num_gaps);
       if (__pyx_t_2) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1381
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1380
  *                     ephr_arr._append(self.eid2symid[self.eda.data.arr[e_sent_start+k]])
  *                 if j < num_gaps:
  *                     indexes.append(sym_setindex(self.category, e_gap_order[j]+1))             # <<<<<<<<<<<<<<
  *                     ephr_arr._append(sym_setindex(self.category, e_gap_order[j]+1))
  *             i = i + step
  */
-        __pyx_t_1 = PyInt_FromLong(__pyx_f_3_sa_sym_setindex(__pyx_v_self->category, ((__pyx_v_e_gap_order[__pyx_v_j]) + 1))); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1381; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_1 = PyInt_FromLong(__pyx_f_3_sa_sym_setindex(__pyx_v_self->category, ((__pyx_v_e_gap_order[__pyx_v_j]) + 1))); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1380; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_1);
-        __pyx_t_10 = PyList_Append(__pyx_v_indexes, __pyx_t_1); if (unlikely(__pyx_t_10 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1381; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_10 = PyList_Append(__pyx_v_indexes, __pyx_t_1); if (unlikely(__pyx_t_10 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1380; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1382
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1381
  *                 if j < num_gaps:
  *                     indexes.append(sym_setindex(self.category, e_gap_order[j]+1))
  *                     ephr_arr._append(sym_setindex(self.category, e_gap_order[j]+1))             # <<<<<<<<<<<<<<
@@ -50903,7 +50955,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
       __pyx_L45:;
     }
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1383
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1382
  *                     indexes.append(sym_setindex(self.category, e_gap_order[j]+1))
  *                     ephr_arr._append(sym_setindex(self.category, e_gap_order[j]+1))
  *             i = i + step             # <<<<<<<<<<<<<<
@@ -50912,7 +50964,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
     __pyx_v_i = (__pyx_v_i + __pyx_v_step);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1384
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1383
  *                     ephr_arr._append(sym_setindex(self.category, e_gap_order[j]+1))
  *             i = i + step
  *             if ephr_arr.len <= self.max_target_length and num_chunks <= self.max_target_chunks:             # <<<<<<<<<<<<<<
@@ -50928,22 +50980,22 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
     }
     if (__pyx_t_7) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1385
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1384
  *             i = i + step
  *             if ephr_arr.len <= self.max_target_length and num_chunks <= self.max_target_chunks:
  *                 result.append((Phrase(ephr_arr),indexes))             # <<<<<<<<<<<<<<
  * 
  *         free(e_gaps1)
  */
-      __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1385; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1384; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_1);
       __Pyx_INCREF(((PyObject *)__pyx_v_ephr_arr));
       PyTuple_SET_ITEM(__pyx_t_1, 0, ((PyObject *)__pyx_v_ephr_arr));
       __Pyx_GIVEREF(((PyObject *)__pyx_v_ephr_arr));
-      __pyx_t_11 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_Phrase)), ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1385; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_11 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_Phrase)), ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1384; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_11);
       __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
-      __pyx_t_1 = PyTuple_New(2); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1385; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_1 = PyTuple_New(2); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1384; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_1);
       PyTuple_SET_ITEM(__pyx_t_1, 0, __pyx_t_11);
       __Pyx_GIVEREF(__pyx_t_11);
@@ -50951,7 +51003,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
       PyTuple_SET_ITEM(__pyx_t_1, 1, ((PyObject *)__pyx_v_indexes));
       __Pyx_GIVEREF(((PyObject *)__pyx_v_indexes));
       __pyx_t_11 = 0;
-      __pyx_t_11 = __Pyx_PyObject_Append(__pyx_v_result, ((PyObject *)__pyx_t_1)); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1385; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_11 = __Pyx_PyObject_Append(__pyx_v_result, ((PyObject *)__pyx_t_1)); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1384; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_11);
       __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
       __Pyx_DECREF(__pyx_t_11); __pyx_t_11 = 0;
@@ -50960,7 +51012,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
     __pyx_L46:;
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1387
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1386
  *                 result.append((Phrase(ephr_arr),indexes))
  * 
  *         free(e_gaps1)             # <<<<<<<<<<<<<<
@@ -50969,7 +51021,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
   free(__pyx_v_e_gaps1);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1388
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1387
  * 
  *         free(e_gaps1)
  *         free(e_gap_order)             # <<<<<<<<<<<<<<
@@ -50978,7 +51030,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
  */
   free(__pyx_v_e_gap_order);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1389
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1388
  *         free(e_gaps1)
  *         free(e_gap_order)
  *         return result             # <<<<<<<<<<<<<<
@@ -51006,7 +51058,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract_phrases(struct _
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1391
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1390
  *         return result
  * 
  *     cdef IntList create_alignments(self, int* sent_links, int num_links, findexes, eindexes):             # <<<<<<<<<<<<<<
@@ -51034,55 +51086,55 @@ static struct __pyx_obj_3_sa_IntList *__pyx_f_3_sa_23HieroCachingRuleFactory_cre
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("create_alignments", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1393
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1392
  *     cdef IntList create_alignments(self, int* sent_links, int num_links, findexes, eindexes):
  *         cdef unsigned i
  *         cdef IntList ret = IntList()             # <<<<<<<<<<<<<<
  *         for i in range(len(findexes)):
  *             s = findexes[i]
  */
-  __pyx_t_1 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_IntList)), ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1393; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_IntList)), ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1392; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __pyx_v_ret = ((struct __pyx_obj_3_sa_IntList *)__pyx_t_1);
   __pyx_t_1 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1394
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1393
  *         cdef unsigned i
  *         cdef IntList ret = IntList()
  *         for i in range(len(findexes)):             # <<<<<<<<<<<<<<
  *             s = findexes[i]
  *             if (s<0):
  */
-  __pyx_t_2 = PyObject_Length(__pyx_v_findexes); if (unlikely(__pyx_t_2 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1394; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Length(__pyx_v_findexes); if (unlikely(__pyx_t_2 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1393; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   for (__pyx_t_3 = 0; __pyx_t_3 < __pyx_t_2; __pyx_t_3+=1) {
     __pyx_v_i = __pyx_t_3;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1395
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1394
  *         cdef IntList ret = IntList()
  *         for i in range(len(findexes)):
  *             s = findexes[i]             # <<<<<<<<<<<<<<
  *             if (s<0):
  *                 continue
  */
-    __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_findexes, __pyx_v_i, sizeof(unsigned int)+1, PyLong_FromUnsignedLong); if (!__pyx_t_1) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1395; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_findexes, __pyx_v_i, sizeof(unsigned int)+1, PyLong_FromUnsignedLong); if (!__pyx_t_1) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1394; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
     __Pyx_XDECREF(__pyx_v_s);
     __pyx_v_s = __pyx_t_1;
     __pyx_t_1 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1396
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1395
  *         for i in range(len(findexes)):
  *             s = findexes[i]
  *             if (s<0):             # <<<<<<<<<<<<<<
  *                 continue
  *             idx = 0
  */
-    __pyx_t_1 = PyObject_RichCompare(__pyx_v_s, __pyx_int_0, Py_LT); __Pyx_XGOTREF(__pyx_t_1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1396; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-    __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1396; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = PyObject_RichCompare(__pyx_v_s, __pyx_int_0, Py_LT); __Pyx_XGOTREF(__pyx_t_1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1395; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1395; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
     if (__pyx_t_4) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1397
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1396
  *             s = findexes[i]
  *             if (s<0):
  *                 continue             # <<<<<<<<<<<<<<
@@ -51094,7 +51146,7 @@ static struct __pyx_obj_3_sa_IntList *__pyx_f_3_sa_23HieroCachingRuleFactory_cre
     }
     __pyx_L5:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1398
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1397
  *             if (s<0):
  *                 continue
  *             idx = 0             # <<<<<<<<<<<<<<
@@ -51105,7 +51157,7 @@ static struct __pyx_obj_3_sa_IntList *__pyx_f_3_sa_23HieroCachingRuleFactory_cre
     __Pyx_XDECREF(__pyx_v_idx);
     __pyx_v_idx = __pyx_int_0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1399
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1398
  *                 continue
  *             idx = 0
  *             while (idx < num_links*2):             # <<<<<<<<<<<<<<
@@ -51113,51 +51165,51 @@ static struct __pyx_obj_3_sa_IntList *__pyx_f_3_sa_23HieroCachingRuleFactory_cre
  *                     j = eindexes.index(sent_links[idx+1])
  */
     while (1) {
-      __pyx_t_1 = PyInt_FromLong((__pyx_v_num_links * 2)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1399; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_1 = PyInt_FromLong((__pyx_v_num_links * 2)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1398; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_1);
-      __pyx_t_5 = PyObject_RichCompare(__pyx_v_idx, __pyx_t_1, Py_LT); __Pyx_XGOTREF(__pyx_t_5); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1399; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_5 = PyObject_RichCompare(__pyx_v_idx, __pyx_t_1, Py_LT); __Pyx_XGOTREF(__pyx_t_5); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1398; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-      __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_5); if (unlikely(__pyx_t_4 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1399; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_5); if (unlikely(__pyx_t_4 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1398; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
       if (!__pyx_t_4) break;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1400
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1399
  *             idx = 0
  *             while (idx < num_links*2):
  *                 if (sent_links[idx] == s):             # <<<<<<<<<<<<<<
  *                     j = eindexes.index(sent_links[idx+1])
  *                     ret.append(i*65536+j)
  */
-      __pyx_t_6 = __Pyx_PyIndex_AsSsize_t(__pyx_v_idx); if (unlikely((__pyx_t_6 == (Py_ssize_t)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1400; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-      __pyx_t_5 = PyInt_FromLong((__pyx_v_sent_links[__pyx_t_6])); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1400; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_6 = __Pyx_PyIndex_AsSsize_t(__pyx_v_idx); if (unlikely((__pyx_t_6 == (Py_ssize_t)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1399; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_5 = PyInt_FromLong((__pyx_v_sent_links[__pyx_t_6])); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1399; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_5);
-      __pyx_t_1 = PyObject_RichCompare(__pyx_t_5, __pyx_v_s, Py_EQ); __Pyx_XGOTREF(__pyx_t_1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1400; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_1 = PyObject_RichCompare(__pyx_t_5, __pyx_v_s, Py_EQ); __Pyx_XGOTREF(__pyx_t_1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1399; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-      __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1400; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1399; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
       if (__pyx_t_4) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1401
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1400
  *             while (idx < num_links*2):
  *                 if (sent_links[idx] == s):
  *                     j = eindexes.index(sent_links[idx+1])             # <<<<<<<<<<<<<<
  *                     ret.append(i*65536+j)
  *                 idx += 2
  */
-        __pyx_t_1 = PyObject_GetAttr(__pyx_v_eindexes, __pyx_n_s__index); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1401; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_1 = PyObject_GetAttr(__pyx_v_eindexes, __pyx_n_s__index); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1400; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_1);
-        __pyx_t_5 = PyNumber_Add(__pyx_v_idx, __pyx_int_1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1401; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_5 = PyNumber_Add(__pyx_v_idx, __pyx_int_1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1400; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_5);
-        __pyx_t_6 = __Pyx_PyIndex_AsSsize_t(__pyx_t_5); if (unlikely((__pyx_t_6 == (Py_ssize_t)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1401; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_6 = __Pyx_PyIndex_AsSsize_t(__pyx_t_5); if (unlikely((__pyx_t_6 == (Py_ssize_t)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1400; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-        __pyx_t_5 = PyInt_FromLong((__pyx_v_sent_links[__pyx_t_6])); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1401; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_5 = PyInt_FromLong((__pyx_v_sent_links[__pyx_t_6])); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1400; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_5);
-        __pyx_t_7 = PyTuple_New(1); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1401; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_7 = PyTuple_New(1); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1400; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_7);
         PyTuple_SET_ITEM(__pyx_t_7, 0, __pyx_t_5);
         __Pyx_GIVEREF(__pyx_t_5);
         __pyx_t_5 = 0;
-        __pyx_t_5 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_7), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1401; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_5 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_7), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1400; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_5);
         __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
         __Pyx_DECREF(((PyObject *)__pyx_t_7)); __pyx_t_7 = 0;
@@ -51165,19 +51217,19 @@ static struct __pyx_obj_3_sa_IntList *__pyx_f_3_sa_23HieroCachingRuleFactory_cre
         __pyx_v_j = __pyx_t_5;
         __pyx_t_5 = 0;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1402
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1401
  *                 if (sent_links[idx] == s):
  *                     j = eindexes.index(sent_links[idx+1])
  *                     ret.append(i*65536+j)             # <<<<<<<<<<<<<<
  *                 idx += 2
  *         return ret
  */
-        __pyx_t_5 = PyInt_FromLong((__pyx_v_i * 65536)); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1402; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_5 = PyInt_FromLong((__pyx_v_i * 65536)); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1401; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_5);
-        __pyx_t_7 = PyNumber_Add(__pyx_t_5, __pyx_v_j); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1402; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_7 = PyNumber_Add(__pyx_t_5, __pyx_v_j); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1401; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_7);
         __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-        __pyx_t_5 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_ret), __pyx_t_7); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1402; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_5 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_ret), __pyx_t_7); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1401; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_5);
         __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
         __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
@@ -51185,14 +51237,14 @@ static struct __pyx_obj_3_sa_IntList *__pyx_f_3_sa_23HieroCachingRuleFactory_cre
       }
       __pyx_L8:;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1403
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1402
  *                     j = eindexes.index(sent_links[idx+1])
  *                     ret.append(i*65536+j)
  *                 idx += 2             # <<<<<<<<<<<<<<
  *         return ret
  * 
  */
-      __pyx_t_5 = PyNumber_InPlaceAdd(__pyx_v_idx, __pyx_int_2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1403; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_5 = PyNumber_InPlaceAdd(__pyx_v_idx, __pyx_int_2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1402; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_5);
       __Pyx_DECREF(__pyx_v_idx);
       __pyx_v_idx = __pyx_t_5;
@@ -51201,7 +51253,7 @@ static struct __pyx_obj_3_sa_IntList *__pyx_f_3_sa_23HieroCachingRuleFactory_cre
     __pyx_L3_continue:;
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1404
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1403
  *                     ret.append(i*65536+j)
  *                 idx += 2
  *         return ret             # <<<<<<<<<<<<<<
@@ -51231,7 +51283,7 @@ static struct __pyx_obj_3_sa_IntList *__pyx_f_3_sa_23HieroCachingRuleFactory_cre
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1406
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1405
  *         return ret
  * 
  *     cdef extract(self, Phrase phrase, Matching* matching, int* chunklen, int num_chunks):             # <<<<<<<<<<<<<<
@@ -51325,19 +51377,19 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("extract", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1419
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1418
  *         cdef reason_for_failure
  * 
  *         fphr_arr = IntList()             # <<<<<<<<<<<<<<
  *         phrase_len = phrase.n
  *         extracts = []
  */
-  __pyx_t_1 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_IntList)), ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1419; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_IntList)), ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1418; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __pyx_v_fphr_arr = ((struct __pyx_obj_3_sa_IntList *)__pyx_t_1);
   __pyx_t_1 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1420
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1419
  * 
  *         fphr_arr = IntList()
  *         phrase_len = phrase.n             # <<<<<<<<<<<<<<
@@ -51346,19 +51398,19 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_phrase_len = __pyx_v_phrase->n;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1421
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1420
  *         fphr_arr = IntList()
  *         phrase_len = phrase.n
  *         extracts = []             # <<<<<<<<<<<<<<
  *         sent_links = self.alignment._get_sent_links(matching.sent_id, &num_links)
  * 
  */
-  __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1421; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1420; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __pyx_v_extracts = ((PyObject *)__pyx_t_1);
   __pyx_t_1 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1422
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1421
  *         phrase_len = phrase.n
  *         extracts = []
  *         sent_links = self.alignment._get_sent_links(matching.sent_id, &num_links)             # <<<<<<<<<<<<<<
@@ -51367,7 +51419,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_sent_links = ((struct __pyx_vtabstruct_3_sa_Alignment *)__pyx_v_self->alignment->__pyx_vtab)->_get_sent_links(__pyx_v_self->alignment, __pyx_v_matching->sent_id, (&__pyx_v_num_links));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1424
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1423
  *         sent_links = self.alignment._get_sent_links(matching.sent_id, &num_links)
  * 
  *         e_sent_start = self.eda.sent_index.arr[matching.sent_id]             # <<<<<<<<<<<<<<
@@ -51376,7 +51428,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_e_sent_start = (__pyx_v_self->eda->sent_index->arr[__pyx_v_matching->sent_id]);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1425
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1424
  * 
  *         e_sent_start = self.eda.sent_index.arr[matching.sent_id]
  *         e_sent_end = self.eda.sent_index.arr[matching.sent_id+1]             # <<<<<<<<<<<<<<
@@ -51385,7 +51437,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_e_sent_end = (__pyx_v_self->eda->sent_index->arr[(__pyx_v_matching->sent_id + 1)]);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1426
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1425
  *         e_sent_start = self.eda.sent_index.arr[matching.sent_id]
  *         e_sent_end = self.eda.sent_index.arr[matching.sent_id+1]
  *         e_sent_len = e_sent_end - e_sent_start - 1             # <<<<<<<<<<<<<<
@@ -51394,7 +51446,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_e_sent_len = ((__pyx_v_e_sent_end - __pyx_v_e_sent_start) - 1);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1427
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1426
  *         e_sent_end = self.eda.sent_index.arr[matching.sent_id+1]
  *         e_sent_len = e_sent_end - e_sent_start - 1
  *         f_sent_start = self.fda.sent_index.arr[matching.sent_id]             # <<<<<<<<<<<<<<
@@ -51403,7 +51455,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_f_sent_start = (__pyx_v_self->fda->sent_index->arr[__pyx_v_matching->sent_id]);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1428
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1427
  *         e_sent_len = e_sent_end - e_sent_start - 1
  *         f_sent_start = self.fda.sent_index.arr[matching.sent_id]
  *         f_sent_end = self.fda.sent_index.arr[matching.sent_id+1]             # <<<<<<<<<<<<<<
@@ -51412,7 +51464,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_f_sent_end = (__pyx_v_self->fda->sent_index->arr[(__pyx_v_matching->sent_id + 1)]);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1429
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1428
  *         f_sent_start = self.fda.sent_index.arr[matching.sent_id]
  *         f_sent_end = self.fda.sent_index.arr[matching.sent_id+1]
  *         f_sent_len = f_sent_end - f_sent_start - 1             # <<<<<<<<<<<<<<
@@ -51421,21 +51473,21 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_f_sent_len = ((__pyx_v_f_sent_end - __pyx_v_f_sent_start) - 1);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1431
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1430
  *         f_sent_len = f_sent_end - f_sent_start - 1
  * 
  *         self.findexes1.reset()             # <<<<<<<<<<<<<<
  *         sofar = 0
  *         for i in range(num_chunks):
  */
-  __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self->findexes1), __pyx_n_s__reset); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1431; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self->findexes1), __pyx_n_s__reset); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1430; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_2 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1431; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1430; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1432
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1431
  * 
  *         self.findexes1.reset()
  *         sofar = 0             # <<<<<<<<<<<<<<
@@ -51445,7 +51497,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
   __Pyx_INCREF(__pyx_int_0);
   __pyx_v_sofar = __pyx_int_0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1433
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1432
  *         self.findexes1.reset()
  *         sofar = 0
  *         for i in range(num_chunks):             # <<<<<<<<<<<<<<
@@ -51456,7 +51508,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
   for (__pyx_t_4 = 0; __pyx_t_4 < __pyx_t_3; __pyx_t_4+=1) {
     __pyx_v_i = __pyx_t_4;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1434
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1433
  *         sofar = 0
  *         for i in range(num_chunks):
  *             for j in range(chunklen[i]):             # <<<<<<<<<<<<<<
@@ -51467,35 +51519,35 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     for (__pyx_t_6 = 0; __pyx_t_6 < __pyx_t_5; __pyx_t_6+=1) {
       __pyx_v_j = __pyx_t_6;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1435
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1434
  *         for i in range(num_chunks):
  *             for j in range(chunklen[i]):
  *                 self.findexes1.append(matching.arr[matching.start+i]+j-f_sent_start);             # <<<<<<<<<<<<<<
  *                 sofar += 1
  *             if (i+1<num_chunks):
  */
-      __pyx_t_2 = PyInt_FromLong((((__pyx_v_matching->arr[(__pyx_v_matching->start + __pyx_v_i)]) + __pyx_v_j) - __pyx_v_f_sent_start)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1435; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyInt_FromLong((((__pyx_v_matching->arr[(__pyx_v_matching->start + __pyx_v_i)]) + __pyx_v_j) - __pyx_v_f_sent_start)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1434; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __pyx_t_1 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes1), __pyx_t_2); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1435; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_1 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes1), __pyx_t_2); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1434; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_1);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
       __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1436
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1435
  *             for j in range(chunklen[i]):
  *                 self.findexes1.append(matching.arr[matching.start+i]+j-f_sent_start);
  *                 sofar += 1             # <<<<<<<<<<<<<<
  *             if (i+1<num_chunks):
  *                 self.findexes1.append(phrase[sofar])
  */
-      __pyx_t_1 = PyNumber_InPlaceAdd(__pyx_v_sofar, __pyx_int_1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1436; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_1 = PyNumber_InPlaceAdd(__pyx_v_sofar, __pyx_int_1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1435; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_1);
       __Pyx_DECREF(__pyx_v_sofar);
       __pyx_v_sofar = __pyx_t_1;
       __pyx_t_1 = 0;
     }
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1437
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1436
  *                 self.findexes1.append(matching.arr[matching.start+i]+j-f_sent_start);
  *                 sofar += 1
  *             if (i+1<num_chunks):             # <<<<<<<<<<<<<<
@@ -51505,28 +51557,28 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     __pyx_t_7 = ((__pyx_v_i + 1) < __pyx_v_num_chunks);
     if (__pyx_t_7) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1438
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1437
  *                 sofar += 1
  *             if (i+1<num_chunks):
  *                 self.findexes1.append(phrase[sofar])             # <<<<<<<<<<<<<<
  *                 sofar += 1
  * 
  */
-      __pyx_t_1 = PyObject_GetItem(((PyObject *)__pyx_v_phrase), __pyx_v_sofar); if (!__pyx_t_1) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1438; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_1 = PyObject_GetItem(((PyObject *)__pyx_v_phrase), __pyx_v_sofar); if (!__pyx_t_1) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1437; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_1);
-      __pyx_t_2 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes1), __pyx_t_1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1438; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes1), __pyx_t_1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1437; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
       __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1439
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1438
  *             if (i+1<num_chunks):
  *                 self.findexes1.append(phrase[sofar])
  *                 sofar += 1             # <<<<<<<<<<<<<<
  * 
  * 
  */
-      __pyx_t_2 = PyNumber_InPlaceAdd(__pyx_v_sofar, __pyx_int_1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1439; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyNumber_InPlaceAdd(__pyx_v_sofar, __pyx_int_1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1438; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
       __Pyx_DECREF(__pyx_v_sofar);
       __pyx_v_sofar = __pyx_t_2;
@@ -51536,7 +51588,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     __pyx_L7:;
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1442
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1441
  * 
  * 
  *         e_links_low = <int*> malloc(e_sent_len*sizeof(int))             # <<<<<<<<<<<<<<
@@ -51545,7 +51597,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_e_links_low = ((int *)malloc((__pyx_v_e_sent_len * (sizeof(int)))));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1443
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1442
  * 
  *         e_links_low = <int*> malloc(e_sent_len*sizeof(int))
  *         e_links_high = <int*> malloc(e_sent_len*sizeof(int))             # <<<<<<<<<<<<<<
@@ -51554,7 +51606,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_e_links_high = ((int *)malloc((__pyx_v_e_sent_len * (sizeof(int)))));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1444
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1443
  *         e_links_low = <int*> malloc(e_sent_len*sizeof(int))
  *         e_links_high = <int*> malloc(e_sent_len*sizeof(int))
  *         f_links_low = <int*> malloc(f_sent_len*sizeof(int))             # <<<<<<<<<<<<<<
@@ -51563,7 +51615,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_f_links_low = ((int *)malloc((__pyx_v_f_sent_len * (sizeof(int)))));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1445
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1444
  *         e_links_high = <int*> malloc(e_sent_len*sizeof(int))
  *         f_links_low = <int*> malloc(f_sent_len*sizeof(int))
  *         f_links_high = <int*> malloc(f_sent_len*sizeof(int))             # <<<<<<<<<<<<<<
@@ -51572,7 +51624,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_f_links_high = ((int *)malloc((__pyx_v_f_sent_len * (sizeof(int)))));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1446
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1445
  *         f_links_low = <int*> malloc(f_sent_len*sizeof(int))
  *         f_links_high = <int*> malloc(f_sent_len*sizeof(int))
  *         f_gap_low = <int*> malloc((num_chunks+1)*sizeof(int))             # <<<<<<<<<<<<<<
@@ -51581,7 +51633,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_f_gap_low = ((int *)malloc(((__pyx_v_num_chunks + 1) * (sizeof(int)))));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1447
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1446
  *         f_links_high = <int*> malloc(f_sent_len*sizeof(int))
  *         f_gap_low = <int*> malloc((num_chunks+1)*sizeof(int))
  *         f_gap_high = <int*> malloc((num_chunks+1)*sizeof(int))             # <<<<<<<<<<<<<<
@@ -51590,7 +51642,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_f_gap_high = ((int *)malloc(((__pyx_v_num_chunks + 1) * (sizeof(int)))));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1448
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1447
  *         f_gap_low = <int*> malloc((num_chunks+1)*sizeof(int))
  *         f_gap_high = <int*> malloc((num_chunks+1)*sizeof(int))
  *         e_gap_low = <int*> malloc((num_chunks+1)*sizeof(int))             # <<<<<<<<<<<<<<
@@ -51599,7 +51651,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_e_gap_low = ((int *)malloc(((__pyx_v_num_chunks + 1) * (sizeof(int)))));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1449
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1448
  *         f_gap_high = <int*> malloc((num_chunks+1)*sizeof(int))
  *         e_gap_low = <int*> malloc((num_chunks+1)*sizeof(int))
  *         e_gap_high = <int*> malloc((num_chunks+1)*sizeof(int))             # <<<<<<<<<<<<<<
@@ -51608,7 +51660,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_e_gap_high = ((int *)malloc(((__pyx_v_num_chunks + 1) * (sizeof(int)))));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1450
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1449
  *         e_gap_low = <int*> malloc((num_chunks+1)*sizeof(int))
  *         e_gap_high = <int*> malloc((num_chunks+1)*sizeof(int))
  *         memset(f_gap_low, 0, (num_chunks+1)*sizeof(int))             # <<<<<<<<<<<<<<
@@ -51617,7 +51669,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   memset(__pyx_v_f_gap_low, 0, ((__pyx_v_num_chunks + 1) * (sizeof(int))));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1451
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1450
  *         e_gap_high = <int*> malloc((num_chunks+1)*sizeof(int))
  *         memset(f_gap_low, 0, (num_chunks+1)*sizeof(int))
  *         memset(f_gap_high, 0, (num_chunks+1)*sizeof(int))             # <<<<<<<<<<<<<<
@@ -51626,7 +51678,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   memset(__pyx_v_f_gap_high, 0, ((__pyx_v_num_chunks + 1) * (sizeof(int))));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1452
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1451
  *         memset(f_gap_low, 0, (num_chunks+1)*sizeof(int))
  *         memset(f_gap_high, 0, (num_chunks+1)*sizeof(int))
  *         memset(e_gap_low, 0, (num_chunks+1)*sizeof(int))             # <<<<<<<<<<<<<<
@@ -51635,7 +51687,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   memset(__pyx_v_e_gap_low, 0, ((__pyx_v_num_chunks + 1) * (sizeof(int))));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1453
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1452
  *         memset(f_gap_high, 0, (num_chunks+1)*sizeof(int))
  *         memset(e_gap_low, 0, (num_chunks+1)*sizeof(int))
  *         memset(e_gap_high, 0, (num_chunks+1)*sizeof(int))             # <<<<<<<<<<<<<<
@@ -51644,7 +51696,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   memset(__pyx_v_e_gap_high, 0, ((__pyx_v_num_chunks + 1) * (sizeof(int))));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1455
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1454
  *         memset(e_gap_high, 0, (num_chunks+1)*sizeof(int))
  * 
  *         reason_for_failure = ""             # <<<<<<<<<<<<<<
@@ -51654,7 +51706,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
   __Pyx_INCREF(((PyObject *)__pyx_kp_s_45));
   __pyx_v_reason_for_failure = ((PyObject *)__pyx_kp_s_45);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1457
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1456
  *         reason_for_failure = ""
  * 
  *         for i from 0 <= i < e_sent_len:             # <<<<<<<<<<<<<<
@@ -51664,7 +51716,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
   __pyx_t_3 = __pyx_v_e_sent_len;
   for (__pyx_v_i = 0; __pyx_v_i < __pyx_t_3; __pyx_v_i++) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1458
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1457
  * 
  *         for i from 0 <= i < e_sent_len:
  *             e_links_low[i] = -1             # <<<<<<<<<<<<<<
@@ -51673,7 +51725,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
     (__pyx_v_e_links_low[__pyx_v_i]) = -1;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1459
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1458
  *         for i from 0 <= i < e_sent_len:
  *             e_links_low[i] = -1
  *             e_links_high[i] = -1             # <<<<<<<<<<<<<<
@@ -51683,7 +51735,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     (__pyx_v_e_links_high[__pyx_v_i]) = -1;
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1460
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1459
  *             e_links_low[i] = -1
  *             e_links_high[i] = -1
  *         for i from 0 <= i < f_sent_len:             # <<<<<<<<<<<<<<
@@ -51693,7 +51745,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
   __pyx_t_3 = __pyx_v_f_sent_len;
   for (__pyx_v_i = 0; __pyx_v_i < __pyx_t_3; __pyx_v_i++) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1461
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1460
  *             e_links_high[i] = -1
  *         for i from 0 <= i < f_sent_len:
  *             f_links_low[i] = -1             # <<<<<<<<<<<<<<
@@ -51702,7 +51754,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
     (__pyx_v_f_links_low[__pyx_v_i]) = -1;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1462
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1461
  *         for i from 0 <= i < f_sent_len:
  *             f_links_low[i] = -1
  *             f_links_high[i] = -1             # <<<<<<<<<<<<<<
@@ -51712,7 +51764,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     (__pyx_v_f_links_high[__pyx_v_i]) = -1;
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1468
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1467
  *         # links that we care about (but then how to look up
  *         # when we want to check something on the e side?)
  *         i = 0             # <<<<<<<<<<<<<<
@@ -51721,7 +51773,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_i = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1469
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1468
  *         # when we want to check something on the e side?)
  *         i = 0
  *         while i < num_links*2:             # <<<<<<<<<<<<<<
@@ -51732,7 +51784,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     __pyx_t_7 = (__pyx_v_i < (__pyx_v_num_links * 2));
     if (!__pyx_t_7) break;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1470
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1469
  *         i = 0
  *         while i < num_links*2:
  *             f_i = sent_links[i]             # <<<<<<<<<<<<<<
@@ -51741,7 +51793,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
     __pyx_v_f_i = (__pyx_v_sent_links[__pyx_v_i]);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1471
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1470
  *         while i < num_links*2:
  *             f_i = sent_links[i]
  *             e_i = sent_links[i+1]             # <<<<<<<<<<<<<<
@@ -51750,7 +51802,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
     __pyx_v_e_i = (__pyx_v_sent_links[(__pyx_v_i + 1)]);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1472
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1471
  *             f_i = sent_links[i]
  *             e_i = sent_links[i+1]
  *             if f_links_low[f_i] == -1 or f_links_low[f_i] > e_i:             # <<<<<<<<<<<<<<
@@ -51766,7 +51818,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     }
     if (__pyx_t_9) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1473
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1472
  *             e_i = sent_links[i+1]
  *             if f_links_low[f_i] == -1 or f_links_low[f_i] > e_i:
  *                 f_links_low[f_i] = e_i             # <<<<<<<<<<<<<<
@@ -51778,7 +51830,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     }
     __pyx_L14:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1474
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1473
  *             if f_links_low[f_i] == -1 or f_links_low[f_i] > e_i:
  *                 f_links_low[f_i] = e_i
  *             if f_links_high[f_i] == -1 or f_links_high[f_i] < e_i + 1:             # <<<<<<<<<<<<<<
@@ -51794,7 +51846,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     }
     if (__pyx_t_8) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1475
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1474
  *                 f_links_low[f_i] = e_i
  *             if f_links_high[f_i] == -1 or f_links_high[f_i] < e_i + 1:
  *                 f_links_high[f_i] = e_i + 1             # <<<<<<<<<<<<<<
@@ -51806,7 +51858,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     }
     __pyx_L15:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1476
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1475
  *             if f_links_high[f_i] == -1 or f_links_high[f_i] < e_i + 1:
  *                 f_links_high[f_i] = e_i + 1
  *             if e_links_low[e_i] == -1 or e_links_low[e_i] > f_i:             # <<<<<<<<<<<<<<
@@ -51822,7 +51874,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     }
     if (__pyx_t_7) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1477
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1476
  *                 f_links_high[f_i] = e_i + 1
  *             if e_links_low[e_i] == -1 or e_links_low[e_i] > f_i:
  *                 e_links_low[e_i] = f_i             # <<<<<<<<<<<<<<
@@ -51834,7 +51886,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     }
     __pyx_L16:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1478
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1477
  *             if e_links_low[e_i] == -1 or e_links_low[e_i] > f_i:
  *                 e_links_low[e_i] = f_i
  *             if e_links_high[e_i] == -1 or e_links_high[e_i] < f_i + 1:             # <<<<<<<<<<<<<<
@@ -51850,7 +51902,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     }
     if (__pyx_t_9) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1479
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1478
  *                 e_links_low[e_i] = f_i
  *             if e_links_high[e_i] == -1 or e_links_high[e_i] < f_i + 1:
  *                 e_links_high[e_i] = f_i + 1             # <<<<<<<<<<<<<<
@@ -51862,7 +51914,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     }
     __pyx_L17:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1480
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1479
  *             if e_links_high[e_i] == -1 or e_links_high[e_i] < f_i + 1:
  *                 e_links_high[e_i] = f_i + 1
  *             i = i + 2             # <<<<<<<<<<<<<<
@@ -51872,19 +51924,19 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     __pyx_v_i = (__pyx_v_i + 2);
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1482
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1481
  *             i = i + 2
  * 
  *         als = []             # <<<<<<<<<<<<<<
  *         for x in range(matching.start,matching.end):
  *             al = (matching.arr[x]-f_sent_start,f_links_low[matching.arr[x]-f_sent_start])
  */
-  __pyx_t_2 = PyList_New(0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1482; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyList_New(0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1481; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_v_als = __pyx_t_2;
   __pyx_t_2 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1483
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1482
  * 
  *         als = []
  *         for x in range(matching.start,matching.end):             # <<<<<<<<<<<<<<
@@ -51895,18 +51947,18 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
   for (__pyx_t_4 = __pyx_v_matching->start; __pyx_t_4 < __pyx_t_3; __pyx_t_4+=1) {
     __pyx_v_x = __pyx_t_4;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1484
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1483
  *         als = []
  *         for x in range(matching.start,matching.end):
  *             al = (matching.arr[x]-f_sent_start,f_links_low[matching.arr[x]-f_sent_start])             # <<<<<<<<<<<<<<
  *             als.append(al)
  *         # check all source-side alignment constraints
  */
-    __pyx_t_2 = PyInt_FromLong(((__pyx_v_matching->arr[__pyx_v_x]) - __pyx_v_f_sent_start)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1484; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyInt_FromLong(((__pyx_v_matching->arr[__pyx_v_x]) - __pyx_v_f_sent_start)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1483; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_1 = PyInt_FromLong((__pyx_v_f_links_low[((__pyx_v_matching->arr[__pyx_v_x]) - __pyx_v_f_sent_start)])); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1484; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = PyInt_FromLong((__pyx_v_f_links_low[((__pyx_v_matching->arr[__pyx_v_x]) - __pyx_v_f_sent_start)])); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1483; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
-    __pyx_t_10 = PyTuple_New(2); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1484; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_10 = PyTuple_New(2); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1483; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_10);
     PyTuple_SET_ITEM(__pyx_t_10, 0, __pyx_t_2);
     __Pyx_GIVEREF(__pyx_t_2);
@@ -51918,17 +51970,17 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     __pyx_v_al = __pyx_t_10;
     __pyx_t_10 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1485
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1484
  *         for x in range(matching.start,matching.end):
  *             al = (matching.arr[x]-f_sent_start,f_links_low[matching.arr[x]-f_sent_start])
  *             als.append(al)             # <<<<<<<<<<<<<<
  *         # check all source-side alignment constraints
  *         met_constraints = 1
  */
-    __pyx_t_11 = PyList_Append(__pyx_v_als, ((PyObject *)__pyx_v_al)); if (unlikely(__pyx_t_11 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1485; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_11 = PyList_Append(__pyx_v_als, ((PyObject *)__pyx_v_al)); if (unlikely(__pyx_t_11 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1484; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1487
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1486
  *             als.append(al)
  *         # check all source-side alignment constraints
  *         met_constraints = 1             # <<<<<<<<<<<<<<
@@ -51937,7 +51989,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_met_constraints = 1;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1488
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1487
  *         # check all source-side alignment constraints
  *         met_constraints = 1
  *         if self.require_aligned_terminal:             # <<<<<<<<<<<<<<
@@ -51946,7 +51998,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   if (__pyx_v_self->require_aligned_terminal) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1489
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1488
  *         met_constraints = 1
  *         if self.require_aligned_terminal:
  *             num_aligned_chunks = 0             # <<<<<<<<<<<<<<
@@ -51955,7 +52007,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
     __pyx_v_num_aligned_chunks = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1490
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1489
  *         if self.require_aligned_terminal:
  *             num_aligned_chunks = 0
  *             for i from 0 <= i < num_chunks:             # <<<<<<<<<<<<<<
@@ -51965,7 +52017,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     __pyx_t_3 = __pyx_v_num_chunks;
     for (__pyx_v_i = 0; __pyx_v_i < __pyx_t_3; __pyx_v_i++) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1491
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1490
  *             num_aligned_chunks = 0
  *             for i from 0 <= i < num_chunks:
  *                 for j from 0 <= j < chunklen[i]:             # <<<<<<<<<<<<<<
@@ -51975,7 +52027,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
       __pyx_t_4 = (__pyx_v_chunklen[__pyx_v_i]);
       for (__pyx_v_j = 0; __pyx_v_j < __pyx_t_4; __pyx_v_j++) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1492
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1491
  *             for i from 0 <= i < num_chunks:
  *                 for j from 0 <= j < chunklen[i]:
  *                     if f_links_low[matching.arr[matching.start+i]+j-f_sent_start] > -1:             # <<<<<<<<<<<<<<
@@ -51985,7 +52037,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
         __pyx_t_9 = ((__pyx_v_f_links_low[(((__pyx_v_matching->arr[(__pyx_v_matching->start + __pyx_v_i)]) + __pyx_v_j) - __pyx_v_f_sent_start)]) > -1);
         if (__pyx_t_9) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1493
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1492
  *                 for j from 0 <= j < chunklen[i]:
  *                     if f_links_low[matching.arr[matching.start+i]+j-f_sent_start] > -1:
  *                         num_aligned_chunks = num_aligned_chunks + 1             # <<<<<<<<<<<<<<
@@ -51994,7 +52046,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
           __pyx_v_num_aligned_chunks = (__pyx_v_num_aligned_chunks + 1);
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1494
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1493
  *                     if f_links_low[matching.arr[matching.start+i]+j-f_sent_start] > -1:
  *                         num_aligned_chunks = num_aligned_chunks + 1
  *                         break             # <<<<<<<<<<<<<<
@@ -52009,7 +52061,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
       __pyx_L24_break:;
     }
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1495
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1494
  *                         num_aligned_chunks = num_aligned_chunks + 1
  *                         break
  *             if num_aligned_chunks == 0:             # <<<<<<<<<<<<<<
@@ -52019,7 +52071,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     __pyx_t_9 = (__pyx_v_num_aligned_chunks == 0);
     if (__pyx_t_9) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1496
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1495
  *                         break
  *             if num_aligned_chunks == 0:
  *                 reason_for_failure = "No aligned terminals"             # <<<<<<<<<<<<<<
@@ -52030,7 +52082,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
       __Pyx_DECREF(__pyx_v_reason_for_failure);
       __pyx_v_reason_for_failure = ((PyObject *)__pyx_kp_s_127);
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1497
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1496
  *             if num_aligned_chunks == 0:
  *                 reason_for_failure = "No aligned terminals"
  *                 met_constraints = 0             # <<<<<<<<<<<<<<
@@ -52042,7 +52094,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     }
     __pyx_L26:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1498
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1497
  *                 reason_for_failure = "No aligned terminals"
  *                 met_constraints = 0
  *             if self.require_aligned_chunks and num_aligned_chunks < num_chunks:             # <<<<<<<<<<<<<<
@@ -52057,7 +52109,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     }
     if (__pyx_t_7) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1499
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1498
  *                 met_constraints = 0
  *             if self.require_aligned_chunks and num_aligned_chunks < num_chunks:
  *                 reason_for_failure = "Unaligned chunk"             # <<<<<<<<<<<<<<
@@ -52068,7 +52120,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
       __Pyx_DECREF(__pyx_v_reason_for_failure);
       __pyx_v_reason_for_failure = ((PyObject *)__pyx_kp_s_128);
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1500
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1499
  *             if self.require_aligned_chunks and num_aligned_chunks < num_chunks:
  *                 reason_for_failure = "Unaligned chunk"
  *                 met_constraints = 0             # <<<<<<<<<<<<<<
@@ -52083,7 +52135,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
   }
   __pyx_L20:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1502
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1501
  *                 met_constraints = 0
  * 
  *         if met_constraints and self.tight_phrases:             # <<<<<<<<<<<<<<
@@ -52097,7 +52149,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
   }
   if (__pyx_t_7) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1504
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1503
  *         if met_constraints and self.tight_phrases:
  *             # outside edge constraints are checked later
  *             for i from 0 <= i < num_chunks-1:             # <<<<<<<<<<<<<<
@@ -52107,7 +52159,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     __pyx_t_12 = (__pyx_v_num_chunks - 1);
     for (__pyx_v_i = 0; __pyx_v_i < __pyx_t_12; __pyx_v_i++) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1505
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1504
  *             # outside edge constraints are checked later
  *             for i from 0 <= i < num_chunks-1:
  *                 if f_links_low[matching.arr[matching.start+i]+chunklen[i]-f_sent_start] == -1:             # <<<<<<<<<<<<<<
@@ -52117,7 +52169,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
       __pyx_t_7 = ((__pyx_v_f_links_low[(((__pyx_v_matching->arr[(__pyx_v_matching->start + __pyx_v_i)]) + (__pyx_v_chunklen[__pyx_v_i])) - __pyx_v_f_sent_start)]) == -1);
       if (__pyx_t_7) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1506
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1505
  *             for i from 0 <= i < num_chunks-1:
  *                 if f_links_low[matching.arr[matching.start+i]+chunklen[i]-f_sent_start] == -1:
  *                     reason_for_failure = "Gaps are not tight phrases"             # <<<<<<<<<<<<<<
@@ -52128,7 +52180,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
         __Pyx_DECREF(__pyx_v_reason_for_failure);
         __pyx_v_reason_for_failure = ((PyObject *)__pyx_kp_s_129);
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1507
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1506
  *                 if f_links_low[matching.arr[matching.start+i]+chunklen[i]-f_sent_start] == -1:
  *                     reason_for_failure = "Gaps are not tight phrases"
  *                     met_constraints = 0             # <<<<<<<<<<<<<<
@@ -52137,7 +52189,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
         __pyx_v_met_constraints = 0;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1508
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1507
  *                     reason_for_failure = "Gaps are not tight phrases"
  *                     met_constraints = 0
  *                     break             # <<<<<<<<<<<<<<
@@ -52149,7 +52201,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
       }
       __pyx_L31:;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1509
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1508
  *                     met_constraints = 0
  *                     break
  *                 if f_links_low[matching.arr[matching.start+i+1]-1-f_sent_start] == -1:             # <<<<<<<<<<<<<<
@@ -52159,7 +52211,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
       __pyx_t_7 = ((__pyx_v_f_links_low[(((__pyx_v_matching->arr[((__pyx_v_matching->start + __pyx_v_i) + 1)]) - 1) - __pyx_v_f_sent_start)]) == -1);
       if (__pyx_t_7) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1510
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1509
  *                     break
  *                 if f_links_low[matching.arr[matching.start+i+1]-1-f_sent_start] == -1:
  *                     reason_for_failure = "Gaps are not tight phrases"             # <<<<<<<<<<<<<<
@@ -52170,7 +52222,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
         __Pyx_DECREF(__pyx_v_reason_for_failure);
         __pyx_v_reason_for_failure = ((PyObject *)__pyx_kp_s_129);
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1511
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1510
  *                 if f_links_low[matching.arr[matching.start+i+1]-1-f_sent_start] == -1:
  *                     reason_for_failure = "Gaps are not tight phrases"
  *                     met_constraints = 0             # <<<<<<<<<<<<<<
@@ -52179,7 +52231,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
         __pyx_v_met_constraints = 0;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1512
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1511
  *                     reason_for_failure = "Gaps are not tight phrases"
  *                     met_constraints = 0
  *                     break             # <<<<<<<<<<<<<<
@@ -52196,7 +52248,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
   }
   __pyx_L28:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1514
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1513
  *                     break
  * 
  *         f_low = matching.arr[matching.start] - f_sent_start             # <<<<<<<<<<<<<<
@@ -52205,7 +52257,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_f_low = ((__pyx_v_matching->arr[__pyx_v_matching->start]) - __pyx_v_f_sent_start);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1515
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1514
  * 
  *         f_low = matching.arr[matching.start] - f_sent_start
  *         f_high = matching.arr[matching.start + matching.size - 1] + chunklen[num_chunks-1] - f_sent_start             # <<<<<<<<<<<<<<
@@ -52214,7 +52266,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   __pyx_v_f_high = (((__pyx_v_matching->arr[((__pyx_v_matching->start + __pyx_v_matching->size) - 1)]) + (__pyx_v_chunklen[(__pyx_v_num_chunks - 1)])) - __pyx_v_f_sent_start);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1516
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1515
  *         f_low = matching.arr[matching.start] - f_sent_start
  *         f_high = matching.arr[matching.start + matching.size - 1] + chunklen[num_chunks-1] - f_sent_start
  *         if met_constraints:             # <<<<<<<<<<<<<<
@@ -52223,17 +52275,17 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   if (__pyx_v_met_constraints) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1518
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1517
  *         if met_constraints:
  * 
  *             if self.find_fixpoint(f_low, f_high, f_links_low, f_links_high, e_links_low, e_links_high,             # <<<<<<<<<<<<<<
  *                                 -1, -1, &e_low, &e_high, &f_back_low, &f_back_high, f_sent_len, e_sent_len,
  *                                 self.train_max_initial_size, self.train_max_initial_size,
  */
-    __pyx_t_10 = PyInt_FromLong(__pyx_v_f_high); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1518; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_10 = PyInt_FromLong(__pyx_v_f_high); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1517; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_10);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1522
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1521
  *                                 self.train_max_initial_size, self.train_max_initial_size,
  *                                 self.train_min_gap_size, 0,
  *                                 self.max_nonterminals - num_chunks + 1, 1, 1, 0, 0):             # <<<<<<<<<<<<<<
@@ -52244,7 +52296,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
     if (__pyx_t_3) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1523
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1522
  *                                 self.train_min_gap_size, 0,
  *                                 self.max_nonterminals - num_chunks + 1, 1, 1, 0, 0):
  *                 gap_error = 0             # <<<<<<<<<<<<<<
@@ -52253,7 +52305,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
       __pyx_v_gap_error = 0;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1524
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1523
  *                                 self.max_nonterminals - num_chunks + 1, 1, 1, 0, 0):
  *                 gap_error = 0
  *                 num_gaps = 0             # <<<<<<<<<<<<<<
@@ -52262,7 +52314,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
       __pyx_v_num_gaps = 0;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1526
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1525
  *                 num_gaps = 0
  * 
  *                 if f_back_low < f_low:             # <<<<<<<<<<<<<<
@@ -52272,7 +52324,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
       __pyx_t_7 = (__pyx_v_f_back_low < __pyx_v_f_low);
       if (__pyx_t_7) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1527
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1526
  * 
  *                 if f_back_low < f_low:
  *                     f_gap_low[0] = f_back_low             # <<<<<<<<<<<<<<
@@ -52281,7 +52333,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
         (__pyx_v_f_gap_low[0]) = __pyx_v_f_back_low;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1528
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1527
  *                 if f_back_low < f_low:
  *                     f_gap_low[0] = f_back_low
  *                     f_gap_high[0] = f_low             # <<<<<<<<<<<<<<
@@ -52290,7 +52342,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
         (__pyx_v_f_gap_high[0]) = __pyx_v_f_low;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1529
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1528
  *                     f_gap_low[0] = f_back_low
  *                     f_gap_high[0] = f_low
  *                     num_gaps = 1             # <<<<<<<<<<<<<<
@@ -52299,7 +52351,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
         __pyx_v_num_gaps = 1;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1530
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1529
  *                     f_gap_high[0] = f_low
  *                     num_gaps = 1
  *                     gap_start = 0             # <<<<<<<<<<<<<<
@@ -52308,7 +52360,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
         __pyx_v_gap_start = 0;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1531
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1530
  *                     num_gaps = 1
  *                     gap_start = 0
  *                     phrase_len = phrase_len+1             # <<<<<<<<<<<<<<
@@ -52317,7 +52369,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
         __pyx_v_phrase_len = (__pyx_v_phrase_len + 1);
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1532
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1531
  *                     gap_start = 0
  *                     phrase_len = phrase_len+1
  *                     if phrase_len > self.max_length:             # <<<<<<<<<<<<<<
@@ -52327,7 +52379,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
         __pyx_t_7 = (__pyx_v_phrase_len > __pyx_v_self->max_length);
         if (__pyx_t_7) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1533
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1532
  *                     phrase_len = phrase_len+1
  *                     if phrase_len > self.max_length:
  *                         gap_error = 1             # <<<<<<<<<<<<<<
@@ -52339,7 +52391,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
         }
         __pyx_L36:;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1534
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1533
  *                     if phrase_len > self.max_length:
  *                         gap_error = 1
  *                     if self.tight_phrases:             # <<<<<<<<<<<<<<
@@ -52348,7 +52400,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
         if (__pyx_v_self->tight_phrases) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1535
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1534
  *                         gap_error = 1
  *                     if self.tight_phrases:
  *                         if f_links_low[f_back_low] == -1 or f_links_low[f_low-1] == -1:             # <<<<<<<<<<<<<<
@@ -52364,7 +52416,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
           }
           if (__pyx_t_8) {
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1536
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1535
  *                     if self.tight_phrases:
  *                         if f_links_low[f_back_low] == -1 or f_links_low[f_low-1] == -1:
  *                             gap_error = 1             # <<<<<<<<<<<<<<
@@ -52373,7 +52425,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             __pyx_v_gap_error = 1;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1537
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1536
  *                         if f_links_low[f_back_low] == -1 or f_links_low[f_low-1] == -1:
  *                             gap_error = 1
  *                             reason_for_failure = "Inside edges of preceding subphrase are not tight"             # <<<<<<<<<<<<<<
@@ -52393,7 +52445,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
       }
       /*else*/ {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1539
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1538
  *                             reason_for_failure = "Inside edges of preceding subphrase are not tight"
  *                 else:
  *                     gap_start = 1             # <<<<<<<<<<<<<<
@@ -52402,7 +52454,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
         __pyx_v_gap_start = 1;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1540
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1539
  *                 else:
  *                     gap_start = 1
  *                     if self.tight_phrases and f_links_low[f_low] == -1:             # <<<<<<<<<<<<<<
@@ -52417,7 +52469,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
         }
         if (__pyx_t_7) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1543
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1542
  *                         # this is not a hard error.    we can't extract this phrase
  *                         # but we still might be able to extract a superphrase
  *                         met_constraints = 0             # <<<<<<<<<<<<<<
@@ -52431,7 +52483,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
       }
       __pyx_L35:;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1545
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1544
  *                         met_constraints = 0
  * 
  *                 for i from 0 <= i < matching.size - 1:             # <<<<<<<<<<<<<<
@@ -52441,7 +52493,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
       __pyx_t_12 = (__pyx_v_matching->size - 1);
       for (__pyx_v_i = 0; __pyx_v_i < __pyx_t_12; __pyx_v_i++) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1546
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1545
  * 
  *                 for i from 0 <= i < matching.size - 1:
  *                     f_gap_low[1+i] = matching.arr[matching.start+i] + chunklen[i] - f_sent_start             # <<<<<<<<<<<<<<
@@ -52450,7 +52502,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
         (__pyx_v_f_gap_low[(1 + __pyx_v_i)]) = (((__pyx_v_matching->arr[(__pyx_v_matching->start + __pyx_v_i)]) + (__pyx_v_chunklen[__pyx_v_i])) - __pyx_v_f_sent_start);
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1547
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1546
  *                 for i from 0 <= i < matching.size - 1:
  *                     f_gap_low[1+i] = matching.arr[matching.start+i] + chunklen[i] - f_sent_start
  *                     f_gap_high[1+i] = matching.arr[matching.start+i+1] - f_sent_start             # <<<<<<<<<<<<<<
@@ -52459,7 +52511,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
         (__pyx_v_f_gap_high[(1 + __pyx_v_i)]) = ((__pyx_v_matching->arr[((__pyx_v_matching->start + __pyx_v_i) + 1)]) - __pyx_v_f_sent_start);
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1548
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1547
  *                     f_gap_low[1+i] = matching.arr[matching.start+i] + chunklen[i] - f_sent_start
  *                     f_gap_high[1+i] = matching.arr[matching.start+i+1] - f_sent_start
  *                     num_gaps = num_gaps + 1             # <<<<<<<<<<<<<<
@@ -52469,7 +52521,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
         __pyx_v_num_gaps = (__pyx_v_num_gaps + 1);
       }
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1550
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1549
  *                     num_gaps = num_gaps + 1
  * 
  *                 if f_high < f_back_high:             # <<<<<<<<<<<<<<
@@ -52479,7 +52531,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
       __pyx_t_7 = (__pyx_v_f_high < __pyx_v_f_back_high);
       if (__pyx_t_7) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1551
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1550
  * 
  *                 if f_high < f_back_high:
  *                     f_gap_low[gap_start+num_gaps] = f_high             # <<<<<<<<<<<<<<
@@ -52488,7 +52540,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
         (__pyx_v_f_gap_low[(__pyx_v_gap_start + __pyx_v_num_gaps)]) = __pyx_v_f_high;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1552
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1551
  *                 if f_high < f_back_high:
  *                     f_gap_low[gap_start+num_gaps] = f_high
  *                     f_gap_high[gap_start+num_gaps] = f_back_high             # <<<<<<<<<<<<<<
@@ -52497,7 +52549,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
         (__pyx_v_f_gap_high[(__pyx_v_gap_start + __pyx_v_num_gaps)]) = __pyx_v_f_back_high;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1553
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1552
  *                     f_gap_low[gap_start+num_gaps] = f_high
  *                     f_gap_high[gap_start+num_gaps] = f_back_high
  *                     num_gaps = num_gaps + 1             # <<<<<<<<<<<<<<
@@ -52506,7 +52558,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
         __pyx_v_num_gaps = (__pyx_v_num_gaps + 1);
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1554
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1553
  *                     f_gap_high[gap_start+num_gaps] = f_back_high
  *                     num_gaps = num_gaps + 1
  *                     phrase_len = phrase_len+1             # <<<<<<<<<<<<<<
@@ -52515,7 +52567,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
         __pyx_v_phrase_len = (__pyx_v_phrase_len + 1);
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1555
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1554
  *                     num_gaps = num_gaps + 1
  *                     phrase_len = phrase_len+1
  *                     if phrase_len > self.max_length:             # <<<<<<<<<<<<<<
@@ -52525,7 +52577,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
         __pyx_t_7 = (__pyx_v_phrase_len > __pyx_v_self->max_length);
         if (__pyx_t_7) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1556
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1555
  *                     phrase_len = phrase_len+1
  *                     if phrase_len > self.max_length:
  *                         gap_error = 1             # <<<<<<<<<<<<<<
@@ -52537,7 +52589,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
         }
         __pyx_L43:;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1557
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1556
  *                     if phrase_len > self.max_length:
  *                         gap_error = 1
  *                     if self.tight_phrases:             # <<<<<<<<<<<<<<
@@ -52546,7 +52598,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
         if (__pyx_v_self->tight_phrases) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1558
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1557
  *                         gap_error = 1
  *                     if self.tight_phrases:
  *                         if f_links_low[f_back_high-1] == -1 or f_links_low[f_high] == -1:             # <<<<<<<<<<<<<<
@@ -52562,7 +52614,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
           }
           if (__pyx_t_9) {
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1559
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1558
  *                     if self.tight_phrases:
  *                         if f_links_low[f_back_high-1] == -1 or f_links_low[f_high] == -1:
  *                             gap_error = 1             # <<<<<<<<<<<<<<
@@ -52571,7 +52623,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             __pyx_v_gap_error = 1;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1560
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1559
  *                         if f_links_low[f_back_high-1] == -1 or f_links_low[f_high] == -1:
  *                             gap_error = 1
  *                             reason_for_failure = "Inside edges of following subphrase are not tight"             # <<<<<<<<<<<<<<
@@ -52591,7 +52643,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
       }
       /*else*/ {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1562
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1561
  *                             reason_for_failure = "Inside edges of following subphrase are not tight"
  *                 else:
  *                     if self.tight_phrases and f_links_low[f_high-1] == -1:             # <<<<<<<<<<<<<<
@@ -52606,7 +52658,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
         }
         if (__pyx_t_7) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1563
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1562
  *                 else:
  *                     if self.tight_phrases and f_links_low[f_high-1] == -1:
  *                         met_constraints = 0             # <<<<<<<<<<<<<<
@@ -52620,7 +52672,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
       }
       __pyx_L42:;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1565
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1564
  *                         met_constraints = 0
  * 
  *                 if gap_error == 0:             # <<<<<<<<<<<<<<
@@ -52630,7 +52682,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
       __pyx_t_7 = (__pyx_v_gap_error == 0);
       if (__pyx_t_7) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1566
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1565
  * 
  *                 if gap_error == 0:
  *                     e_word_count = e_high - e_low             # <<<<<<<<<<<<<<
@@ -52639,7 +52691,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
         __pyx_v_e_word_count = (__pyx_v_e_high - __pyx_v_e_low);
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1567
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1566
  *                 if gap_error == 0:
  *                     e_word_count = e_high - e_low
  *                     for i from 0 <= i < num_gaps: # check integrity of subphrases             # <<<<<<<<<<<<<<
@@ -52649,17 +52701,17 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
         __pyx_t_3 = __pyx_v_num_gaps;
         for (__pyx_v_i = 0; __pyx_v_i < __pyx_t_3; __pyx_v_i++) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1568
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1567
  *                     e_word_count = e_high - e_low
  *                     for i from 0 <= i < num_gaps: # check integrity of subphrases
  *                         if self.find_fixpoint(f_gap_low[gap_start+i], f_gap_high[gap_start+i],             # <<<<<<<<<<<<<<
  *                                             f_links_low, f_links_high, e_links_low, e_links_high,
  *                                             -1, -1, e_gap_low+gap_start+i, e_gap_high+gap_start+i,
  */
-          __pyx_t_10 = PyInt_FromLong((__pyx_v_f_gap_high[(__pyx_v_gap_start + __pyx_v_i)])); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1568; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_10 = PyInt_FromLong((__pyx_v_f_gap_high[(__pyx_v_gap_start + __pyx_v_i)])); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1567; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_10);
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1573
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1572
  *                                             f_gap_low+gap_start+i, f_gap_high+gap_start+i,
  *                                             f_sent_len, e_sent_len,
  *                                             self.train_max_initial_size, self.train_max_initial_size,             # <<<<<<<<<<<<<<
@@ -52670,7 +52722,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
           __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
           if (__pyx_t_7) {
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1575
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1574
  *                                             self.train_max_initial_size, self.train_max_initial_size,
  *                                             0, 0, 0, 0, 0, 0, 0) == 0:
  *                             gap_error = 1             # <<<<<<<<<<<<<<
@@ -52679,18 +52731,18 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             __pyx_v_gap_error = 1;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1576
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1575
  *                                             0, 0, 0, 0, 0, 0, 0) == 0:
  *                             gap_error = 1
  *                             reason_for_failure = "Subphrase [%d, %d] failed integrity check" % (f_gap_low[gap_start+i], f_gap_high[gap_start+i])             # <<<<<<<<<<<<<<
  *                             break
  * 
  */
-            __pyx_t_10 = PyInt_FromLong((__pyx_v_f_gap_low[(__pyx_v_gap_start + __pyx_v_i)])); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1576; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_10 = PyInt_FromLong((__pyx_v_f_gap_low[(__pyx_v_gap_start + __pyx_v_i)])); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1575; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_10);
-            __pyx_t_1 = PyInt_FromLong((__pyx_v_f_gap_high[(__pyx_v_gap_start + __pyx_v_i)])); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1576; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_1 = PyInt_FromLong((__pyx_v_f_gap_high[(__pyx_v_gap_start + __pyx_v_i)])); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1575; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_1);
-            __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1576; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1575; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_2);
             PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_t_10);
             __Pyx_GIVEREF(__pyx_t_10);
@@ -52698,14 +52750,14 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             __Pyx_GIVEREF(__pyx_t_1);
             __pyx_t_10 = 0;
             __pyx_t_1 = 0;
-            __pyx_t_1 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_132), ((PyObject *)__pyx_t_2)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1576; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_1 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_132), ((PyObject *)__pyx_t_2)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1575; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(((PyObject *)__pyx_t_1));
             __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
             __Pyx_DECREF(__pyx_v_reason_for_failure);
             __pyx_v_reason_for_failure = ((PyObject *)__pyx_t_1);
             __pyx_t_1 = 0;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1577
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1576
  *                             gap_error = 1
  *                             reason_for_failure = "Subphrase [%d, %d] failed integrity check" % (f_gap_low[gap_start+i], f_gap_high[gap_start+i])
  *                             break             # <<<<<<<<<<<<<<
@@ -52722,7 +52774,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
       }
       __pyx_L47:;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1579
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1578
  *                             break
  * 
  *                 if gap_error == 0:             # <<<<<<<<<<<<<<
@@ -52732,7 +52784,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
       __pyx_t_7 = (__pyx_v_gap_error == 0);
       if (__pyx_t_7) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1580
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1579
  * 
  *                 if gap_error == 0:
  *                     i = 1             # <<<<<<<<<<<<<<
@@ -52741,21 +52793,21 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
         __pyx_v_i = 1;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1581
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1580
  *                 if gap_error == 0:
  *                     i = 1
  *                     self.findexes.reset()             # <<<<<<<<<<<<<<
  *                     if f_back_low < f_low:
  *                         fphr_arr._append(sym_setindex(self.category, i))
  */
-        __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self->findexes), __pyx_n_s__reset); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1581; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self->findexes), __pyx_n_s__reset); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1580; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_1);
-        __pyx_t_2 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1581; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_2 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1580; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_2);
         __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
         __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1582
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1581
  *                     i = 1
  *                     self.findexes.reset()
  *                     if f_back_low < f_low:             # <<<<<<<<<<<<<<
@@ -52765,7 +52817,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
         __pyx_t_7 = (__pyx_v_f_back_low < __pyx_v_f_low);
         if (__pyx_t_7) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1583
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1582
  *                     self.findexes.reset()
  *                     if f_back_low < f_low:
  *                         fphr_arr._append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
@@ -52774,7 +52826,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
           ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_fphr_arr->__pyx_vtab)->_append(__pyx_v_fphr_arr, __pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i));
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1584
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1583
  *                     if f_back_low < f_low:
  *                         fphr_arr._append(sym_setindex(self.category, i))
  *                         i = i+1             # <<<<<<<<<<<<<<
@@ -52783,16 +52835,16 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
           __pyx_v_i = (__pyx_v_i + 1);
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1585
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1584
  *                         fphr_arr._append(sym_setindex(self.category, i))
  *                         i = i+1
  *                         self.findexes.append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
  *                     self.findexes.extend(self.findexes1)
  *                     for j from 0 <= j < phrase.n:
  */
-          __pyx_t_2 = PyInt_FromLong(__pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1585; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_2 = PyInt_FromLong(__pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1584; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_2);
-          __pyx_t_1 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes), __pyx_t_2); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1585; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_1 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes), __pyx_t_2); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1584; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_1);
           __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
           __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
@@ -52800,27 +52852,27 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
         }
         __pyx_L52:;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1586
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1585
  *                         i = i+1
  *                         self.findexes.append(sym_setindex(self.category, i))
  *                     self.findexes.extend(self.findexes1)             # <<<<<<<<<<<<<<
  *                     for j from 0 <= j < phrase.n:
  *                         if sym_isvar(phrase.syms[j]):
  */
-        __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self->findexes), __pyx_n_s__extend); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1586; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self->findexes), __pyx_n_s__extend); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1585; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_1);
-        __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1586; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1585; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_2);
         __Pyx_INCREF(((PyObject *)__pyx_v_self->findexes1));
         PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_v_self->findexes1));
         __Pyx_GIVEREF(((PyObject *)__pyx_v_self->findexes1));
-        __pyx_t_10 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1586; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_10 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1585; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_10);
         __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
         __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
         __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1587
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1586
  *                         self.findexes.append(sym_setindex(self.category, i))
  *                     self.findexes.extend(self.findexes1)
  *                     for j from 0 <= j < phrase.n:             # <<<<<<<<<<<<<<
@@ -52830,7 +52882,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
         __pyx_t_3 = __pyx_v_phrase->n;
         for (__pyx_v_j = 0; __pyx_v_j < __pyx_t_3; __pyx_v_j++) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1588
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1587
  *                     self.findexes.extend(self.findexes1)
  *                     for j from 0 <= j < phrase.n:
  *                         if sym_isvar(phrase.syms[j]):             # <<<<<<<<<<<<<<
@@ -52840,7 +52892,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
           __pyx_t_4 = __pyx_f_3_sa_sym_isvar((__pyx_v_phrase->syms[__pyx_v_j]));
           if (__pyx_t_4) {
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1589
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1588
  *                     for j from 0 <= j < phrase.n:
  *                         if sym_isvar(phrase.syms[j]):
  *                             fphr_arr._append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
@@ -52849,7 +52901,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_fphr_arr->__pyx_vtab)->_append(__pyx_v_fphr_arr, __pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i));
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1590
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1589
  *                         if sym_isvar(phrase.syms[j]):
  *                             fphr_arr._append(sym_setindex(self.category, i))
  *                             i = i + 1             # <<<<<<<<<<<<<<
@@ -52861,7 +52913,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
           }
           /*else*/ {
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1592
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1591
  *                             i = i + 1
  *                         else:
  *                             fphr_arr._append(phrase.syms[j])             # <<<<<<<<<<<<<<
@@ -52873,7 +52925,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
           __pyx_L55:;
         }
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1593
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1592
  *                         else:
  *                             fphr_arr._append(phrase.syms[j])
  *                     if f_back_high > f_high:             # <<<<<<<<<<<<<<
@@ -52883,7 +52935,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
         __pyx_t_7 = (__pyx_v_f_back_high > __pyx_v_f_high);
         if (__pyx_t_7) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1594
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1593
  *                             fphr_arr._append(phrase.syms[j])
  *                     if f_back_high > f_high:
  *                         fphr_arr._append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
@@ -52892,16 +52944,16 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
           ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_fphr_arr->__pyx_vtab)->_append(__pyx_v_fphr_arr, __pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i));
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1595
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1594
  *                     if f_back_high > f_high:
  *                         fphr_arr._append(sym_setindex(self.category, i))
  *                         self.findexes.append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
  * 
  *                     fphr = Phrase(fphr_arr)
  */
-          __pyx_t_10 = PyInt_FromLong(__pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i)); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1595; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_10 = PyInt_FromLong(__pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i)); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1594; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_10);
-          __pyx_t_2 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes), __pyx_t_10); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1595; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_2 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes), __pyx_t_10); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1594; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_2);
           __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
           __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
@@ -52909,25 +52961,25 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
         }
         __pyx_L56:;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1597
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1596
  *                         self.findexes.append(sym_setindex(self.category, i))
  * 
  *                     fphr = Phrase(fphr_arr)             # <<<<<<<<<<<<<<
  *                     if met_constraints:
  *                         phrase_list = self.extract_phrases(e_low, e_high, e_gap_low + gap_start, e_gap_high + gap_start, e_links_low, num_gaps,
  */
-        __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1597; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1596; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_2);
         __Pyx_INCREF(((PyObject *)__pyx_v_fphr_arr));
         PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_v_fphr_arr));
         __Pyx_GIVEREF(((PyObject *)__pyx_v_fphr_arr));
-        __pyx_t_10 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_Phrase)), ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1597; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_10 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_Phrase)), ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1596; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_10);
         __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
         __pyx_v_fphr = ((struct __pyx_obj_3_sa_Phrase *)__pyx_t_10);
         __pyx_t_10 = 0;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1598
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1597
  * 
  *                     fphr = Phrase(fphr_arr)
  *                     if met_constraints:             # <<<<<<<<<<<<<<
@@ -52936,47 +52988,47 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
         if (__pyx_v_met_constraints) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1601
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1600
  *                         phrase_list = self.extract_phrases(e_low, e_high, e_gap_low + gap_start, e_gap_high + gap_start, e_links_low, num_gaps,
  *                                             f_back_low, f_back_high, f_gap_low + gap_start, f_gap_high + gap_start, f_links_low,
  *                                             matching.sent_id, e_sent_len, e_sent_start)             # <<<<<<<<<<<<<<
  *                         if len(phrase_list) > 0:
  *                             pair_count = 1.0 / len(phrase_list)
  */
-          __pyx_t_10 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->extract_phrases(__pyx_v_self, __pyx_v_e_low, __pyx_v_e_high, (__pyx_v_e_gap_low + __pyx_v_gap_start), (__pyx_v_e_gap_high + __pyx_v_gap_start), __pyx_v_e_links_low, __pyx_v_num_gaps, __pyx_v_f_back_low, __pyx_v_f_back_high, (__pyx_v_f_gap_low + __pyx_v_gap_start), (__pyx_v_f_gap_high + __pyx_v_gap_start), __pyx_v_f_links_low, __pyx_v_matching->sent_id, __pyx_v_e_sent_len, __pyx_v_e_sent_start); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1599; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_10 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->extract_phrases(__pyx_v_self, __pyx_v_e_low, __pyx_v_e_high, (__pyx_v_e_gap_low + __pyx_v_gap_start), (__pyx_v_e_gap_high + __pyx_v_gap_start), __pyx_v_e_links_low, __pyx_v_num_gaps, __pyx_v_f_back_low, __pyx_v_f_back_high, (__pyx_v_f_gap_low + __pyx_v_gap_start), (__pyx_v_f_gap_high + __pyx_v_gap_start), __pyx_v_f_links_low, __pyx_v_matching->sent_id, __pyx_v_e_sent_len, __pyx_v_e_sent_start); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1598; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __Pyx_GOTREF(__pyx_t_10);
           __pyx_v_phrase_list = __pyx_t_10;
           __pyx_t_10 = 0;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1602
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1601
  *                                             f_back_low, f_back_high, f_gap_low + gap_start, f_gap_high + gap_start, f_links_low,
  *                                             matching.sent_id, e_sent_len, e_sent_start)
  *                         if len(phrase_list) > 0:             # <<<<<<<<<<<<<<
  *                             pair_count = 1.0 / len(phrase_list)
  *                         else:
  */
-          __pyx_t_13 = PyObject_Length(__pyx_v_phrase_list); if (unlikely(__pyx_t_13 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1602; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_13 = PyObject_Length(__pyx_v_phrase_list); if (unlikely(__pyx_t_13 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1601; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           __pyx_t_7 = (__pyx_t_13 > 0);
           if (__pyx_t_7) {
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1603
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1602
  *                                             matching.sent_id, e_sent_len, e_sent_start)
  *                         if len(phrase_list) > 0:
  *                             pair_count = 1.0 / len(phrase_list)             # <<<<<<<<<<<<<<
  *                         else:
  *                             pair_count = 0
  */
-            __pyx_t_13 = PyObject_Length(__pyx_v_phrase_list); if (unlikely(__pyx_t_13 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1603; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_13 = PyObject_Length(__pyx_v_phrase_list); if (unlikely(__pyx_t_13 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1602; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             if (unlikely(__pyx_t_13 == 0)) {
               PyErr_Format(PyExc_ZeroDivisionError, "float division");
-              {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1603; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1602; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             }
             __pyx_v_pair_count = (1.0 / __pyx_t_13);
             goto __pyx_L58;
           }
           /*else*/ {
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1605
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1604
  *                             pair_count = 1.0 / len(phrase_list)
  *                         else:
  *                             pair_count = 0             # <<<<<<<<<<<<<<
@@ -52985,22 +53037,22 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             __pyx_v_pair_count = 0.0;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1606
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1605
  *                         else:
  *                             pair_count = 0
  *                             reason_for_failure = "Didn't extract anything from [%d, %d] -> [%d, %d]" % (f_back_low, f_back_high, e_low, e_high)             # <<<<<<<<<<<<<<
  *                         for (phrase2,eindexes) in phrase_list:
  *                             als1 = self.create_alignments(sent_links,num_links,self.findexes,eindexes)
  */
-            __pyx_t_10 = PyInt_FromLong(__pyx_v_f_back_low); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1606; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_10 = PyInt_FromLong(__pyx_v_f_back_low); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1605; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_10);
-            __pyx_t_2 = PyInt_FromLong(__pyx_v_f_back_high); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1606; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_2 = PyInt_FromLong(__pyx_v_f_back_high); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1605; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_2);
-            __pyx_t_1 = PyInt_FromLong(__pyx_v_e_low); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1606; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_1 = PyInt_FromLong(__pyx_v_e_low); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1605; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_1);
-            __pyx_t_14 = PyInt_FromLong(__pyx_v_e_high); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1606; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_14 = PyInt_FromLong(__pyx_v_e_high); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1605; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_14);
-            __pyx_t_15 = PyTuple_New(4); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1606; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_15 = PyTuple_New(4); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1605; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_15);
             PyTuple_SET_ITEM(__pyx_t_15, 0, __pyx_t_10);
             __Pyx_GIVEREF(__pyx_t_10);
@@ -53014,7 +53066,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             __pyx_t_2 = 0;
             __pyx_t_1 = 0;
             __pyx_t_14 = 0;
-            __pyx_t_14 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_133), ((PyObject *)__pyx_t_15)); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1606; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_14 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_133), ((PyObject *)__pyx_t_15)); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1605; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(((PyObject *)__pyx_t_14));
             __Pyx_DECREF(((PyObject *)__pyx_t_15)); __pyx_t_15 = 0;
             __Pyx_DECREF(__pyx_v_reason_for_failure);
@@ -53023,7 +53075,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
           }
           __pyx_L58:;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1607
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1606
  *                             pair_count = 0
  *                             reason_for_failure = "Didn't extract anything from [%d, %d] -> [%d, %d]" % (f_back_low, f_back_high, e_low, e_high)
  *                         for (phrase2,eindexes) in phrase_list:             # <<<<<<<<<<<<<<
@@ -53034,7 +53086,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             __pyx_t_14 = __pyx_v_phrase_list; __Pyx_INCREF(__pyx_t_14); __pyx_t_13 = 0;
             __pyx_t_16 = NULL;
           } else {
-            __pyx_t_13 = -1; __pyx_t_14 = PyObject_GetIter(__pyx_v_phrase_list); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1607; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_13 = -1; __pyx_t_14 = PyObject_GetIter(__pyx_v_phrase_list); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1606; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_14);
             __pyx_t_16 = Py_TYPE(__pyx_t_14)->tp_iternext;
           }
@@ -53042,23 +53094,23 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             if (!__pyx_t_16 && PyList_CheckExact(__pyx_t_14)) {
               if (__pyx_t_13 >= PyList_GET_SIZE(__pyx_t_14)) break;
               #if CYTHON_COMPILING_IN_CPYTHON
-              __pyx_t_15 = PyList_GET_ITEM(__pyx_t_14, __pyx_t_13); __Pyx_INCREF(__pyx_t_15); __pyx_t_13++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1607; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_15 = PyList_GET_ITEM(__pyx_t_14, __pyx_t_13); __Pyx_INCREF(__pyx_t_15); __pyx_t_13++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1606; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               #else
-              __pyx_t_15 = PySequence_ITEM(__pyx_t_14, __pyx_t_13); __pyx_t_13++; if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1607; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_15 = PySequence_ITEM(__pyx_t_14, __pyx_t_13); __pyx_t_13++; if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1606; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               #endif
             } else if (!__pyx_t_16 && PyTuple_CheckExact(__pyx_t_14)) {
               if (__pyx_t_13 >= PyTuple_GET_SIZE(__pyx_t_14)) break;
               #if CYTHON_COMPILING_IN_CPYTHON
-              __pyx_t_15 = PyTuple_GET_ITEM(__pyx_t_14, __pyx_t_13); __Pyx_INCREF(__pyx_t_15); __pyx_t_13++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1607; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_15 = PyTuple_GET_ITEM(__pyx_t_14, __pyx_t_13); __Pyx_INCREF(__pyx_t_15); __pyx_t_13++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1606; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               #else
-              __pyx_t_15 = PySequence_ITEM(__pyx_t_14, __pyx_t_13); __pyx_t_13++; if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1607; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_15 = PySequence_ITEM(__pyx_t_14, __pyx_t_13); __pyx_t_13++; if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1606; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               #endif
             } else {
               __pyx_t_15 = __pyx_t_16(__pyx_t_14);
               if (unlikely(!__pyx_t_15)) {
                 if (PyErr_Occurred()) {
                   if (likely(PyErr_ExceptionMatches(PyExc_StopIteration))) PyErr_Clear();
-                  else {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1607; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  else {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1606; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 }
                 break;
               }
@@ -53074,7 +53126,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
               if (unlikely(size != 2)) {
                 if (size > 2) __Pyx_RaiseTooManyValuesError(2);
                 else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size);
-                {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1607; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1606; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               }
               #if CYTHON_COMPILING_IN_CPYTHON
               if (likely(PyTuple_CheckExact(sequence))) {
@@ -53087,14 +53139,14 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
               __Pyx_INCREF(__pyx_t_1);
               __Pyx_INCREF(__pyx_t_2);
               #else
-              __pyx_t_1 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1607; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-              __pyx_t_2 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1607; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_1 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1606; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_2 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1606; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               #endif
               __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
             } else
             {
               Py_ssize_t index = -1;
-              __pyx_t_10 = PyObject_GetIter(__pyx_t_15); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1607; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_10 = PyObject_GetIter(__pyx_t_15); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1606; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_10);
               __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
               __pyx_t_17 = Py_TYPE(__pyx_t_10)->tp_iternext;
@@ -53102,7 +53154,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
               __Pyx_GOTREF(__pyx_t_1);
               index = 1; __pyx_t_2 = __pyx_t_17(__pyx_t_10); if (unlikely(!__pyx_t_2)) goto __pyx_L61_unpacking_failed;
               __Pyx_GOTREF(__pyx_t_2);
-              if (__Pyx_IternextUnpackEndCheck(__pyx_t_17(__pyx_t_10), 2) < 0) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1607; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              if (__Pyx_IternextUnpackEndCheck(__pyx_t_17(__pyx_t_10), 2) < 0) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1606; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __pyx_t_17 = NULL;
               __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
               goto __pyx_L62_unpacking_done;
@@ -53110,7 +53162,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
               __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
               __pyx_t_17 = NULL;
               if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index);
-              {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1607; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1606; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __pyx_L62_unpacking_done:;
             }
             __Pyx_XDECREF(__pyx_v_phrase2);
@@ -53120,7 +53172,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             __pyx_v_eindexes = __pyx_t_2;
             __pyx_t_2 = 0;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1608
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1607
  *                             reason_for_failure = "Didn't extract anything from [%d, %d] -> [%d, %d]" % (f_back_low, f_back_high, e_low, e_high)
  *                         for (phrase2,eindexes) in phrase_list:
  *                             als1 = self.create_alignments(sent_links,num_links,self.findexes,eindexes)             # <<<<<<<<<<<<<<
@@ -53129,31 +53181,31 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             __pyx_t_15 = ((PyObject *)__pyx_v_self->findexes);
             __Pyx_INCREF(__pyx_t_15);
-            __pyx_t_2 = ((PyObject *)((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->create_alignments(__pyx_v_self, __pyx_v_sent_links, __pyx_v_num_links, __pyx_t_15, __pyx_v_eindexes)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1608; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_2 = ((PyObject *)((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->create_alignments(__pyx_v_self, __pyx_v_sent_links, __pyx_v_num_links, __pyx_t_15, __pyx_v_eindexes)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1607; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_2);
             __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
             __Pyx_XDECREF(((PyObject *)__pyx_v_als1));
             __pyx_v_als1 = ((struct __pyx_obj_3_sa_IntList *)__pyx_t_2);
             __pyx_t_2 = 0;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1609
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1608
  *                         for (phrase2,eindexes) in phrase_list:
  *                             als1 = self.create_alignments(sent_links,num_links,self.findexes,eindexes)
  *                             extracts.append((fphr, phrase2, pair_count, tuple(als1)))             # <<<<<<<<<<<<<<
  * 
  *                     if (num_gaps < self.max_nonterminals and
  */
-            __pyx_t_2 = PyFloat_FromDouble(__pyx_v_pair_count); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1609; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_2 = PyFloat_FromDouble(__pyx_v_pair_count); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1608; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_2);
-            __pyx_t_15 = PyTuple_New(1); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1609; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_15 = PyTuple_New(1); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1608; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_15);
             __Pyx_INCREF(((PyObject *)__pyx_v_als1));
             PyTuple_SET_ITEM(__pyx_t_15, 0, ((PyObject *)__pyx_v_als1));
             __Pyx_GIVEREF(((PyObject *)__pyx_v_als1));
-            __pyx_t_1 = PyObject_Call(((PyObject *)((PyObject*)(&PyTuple_Type))), ((PyObject *)__pyx_t_15), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1609; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_1 = PyObject_Call(((PyObject *)((PyObject*)(&PyTuple_Type))), ((PyObject *)__pyx_t_15), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1608; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_1);
             __Pyx_DECREF(((PyObject *)__pyx_t_15)); __pyx_t_15 = 0;
-            __pyx_t_15 = PyTuple_New(4); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1609; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_15 = PyTuple_New(4); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1608; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_15);
             __Pyx_INCREF(((PyObject *)__pyx_v_fphr));
             PyTuple_SET_ITEM(__pyx_t_15, 0, ((PyObject *)__pyx_v_fphr));
@@ -53167,7 +53219,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             __Pyx_GIVEREF(__pyx_t_1);
             __pyx_t_2 = 0;
             __pyx_t_1 = 0;
-            __pyx_t_1 = __Pyx_PyObject_Append(__pyx_v_extracts, ((PyObject *)__pyx_t_15)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1609; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_1 = __Pyx_PyObject_Append(__pyx_v_extracts, ((PyObject *)__pyx_t_15)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1608; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_1);
             __Pyx_DECREF(((PyObject *)__pyx_t_15)); __pyx_t_15 = 0;
             __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
@@ -53177,7 +53229,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
         }
         __pyx_L57:;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1611
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1610
  *                             extracts.append((fphr, phrase2, pair_count, tuple(als1)))
  * 
  *                     if (num_gaps < self.max_nonterminals and             # <<<<<<<<<<<<<<
@@ -53187,7 +53239,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
         __pyx_t_7 = (__pyx_v_num_gaps < __pyx_v_self->max_nonterminals);
         if (__pyx_t_7) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1612
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1611
  * 
  *                     if (num_gaps < self.max_nonterminals and
  *                         phrase_len < self.max_length and             # <<<<<<<<<<<<<<
@@ -53197,7 +53249,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
           __pyx_t_9 = (__pyx_v_phrase_len < __pyx_v_self->max_length);
           if (__pyx_t_9) {
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1613
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1612
  *                     if (num_gaps < self.max_nonterminals and
  *                         phrase_len < self.max_length and
  *                         f_back_high - f_back_low + self.train_min_gap_size <= self.train_max_initial_size):             # <<<<<<<<<<<<<<
@@ -53215,7 +53267,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
         }
         if (__pyx_t_9) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1614
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1613
  *                         phrase_len < self.max_length and
  *                         f_back_high - f_back_low + self.train_min_gap_size <= self.train_max_initial_size):
  *                         if (f_back_low == f_low and             # <<<<<<<<<<<<<<
@@ -53225,7 +53277,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
           __pyx_t_9 = (__pyx_v_f_back_low == __pyx_v_f_low);
           if (__pyx_t_9) {
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1615
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1614
  *                         f_back_high - f_back_low + self.train_min_gap_size <= self.train_max_initial_size):
  *                         if (f_back_low == f_low and
  *                                 f_low >= self.train_min_gap_size and             # <<<<<<<<<<<<<<
@@ -53235,7 +53287,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             __pyx_t_7 = (__pyx_v_f_low >= __pyx_v_self->train_min_gap_size);
             if (__pyx_t_7) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1616
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1615
  *                         if (f_back_low == f_low and
  *                                 f_low >= self.train_min_gap_size and
  *                                 ((not self.tight_phrases) or (f_links_low[f_low-1] != -1 and f_links_low[f_back_high-1] != -1))):             # <<<<<<<<<<<<<<
@@ -53265,7 +53317,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
           }
           if (__pyx_t_7) {
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1617
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1616
  *                                 f_low >= self.train_min_gap_size and
  *                                 ((not self.tight_phrases) or (f_links_low[f_low-1] != -1 and f_links_low[f_back_high-1] != -1))):
  *                             f_x_low = f_low-self.train_min_gap_size             # <<<<<<<<<<<<<<
@@ -53274,7 +53326,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             __pyx_v_f_x_low = (__pyx_v_f_low - __pyx_v_self->train_min_gap_size);
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1618
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1617
  *                                 ((not self.tight_phrases) or (f_links_low[f_low-1] != -1 and f_links_low[f_back_high-1] != -1))):
  *                             f_x_low = f_low-self.train_min_gap_size
  *                             met_constraints = 1             # <<<<<<<<<<<<<<
@@ -53283,7 +53335,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             __pyx_v_met_constraints = 1;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1619
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1618
  *                             f_x_low = f_low-self.train_min_gap_size
  *                             met_constraints = 1
  *                             if self.tight_phrases:             # <<<<<<<<<<<<<<
@@ -53292,7 +53344,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             if (__pyx_v_self->tight_phrases) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1620
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1619
  *                             met_constraints = 1
  *                             if self.tight_phrases:
  *                                 while f_x_low >= 0 and f_links_low[f_x_low] == -1:             # <<<<<<<<<<<<<<
@@ -53309,7 +53361,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 }
                 if (!__pyx_t_18) break;
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1621
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1620
  *                             if self.tight_phrases:
  *                                 while f_x_low >= 0 and f_links_low[f_x_low] == -1:
  *                                     f_x_low = f_x_low - 1             # <<<<<<<<<<<<<<
@@ -53322,7 +53374,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             }
             __pyx_L65:;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1622
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1621
  *                                 while f_x_low >= 0 and f_links_low[f_x_low] == -1:
  *                                     f_x_low = f_x_low - 1
  *                             if f_x_low < 0 or f_back_high - f_x_low > self.train_max_initial_size:             # <<<<<<<<<<<<<<
@@ -53338,7 +53390,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             }
             if (__pyx_t_9) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1623
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1622
  *                                     f_x_low = f_x_low - 1
  *                             if f_x_low < 0 or f_back_high - f_x_low > self.train_max_initial_size:
  *                                 met_constraints = 0             # <<<<<<<<<<<<<<
@@ -53350,7 +53402,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             }
             __pyx_L68:;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1625
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1624
  *                                 met_constraints = 0
  * 
  *                             if (met_constraints and             # <<<<<<<<<<<<<<
@@ -53359,17 +53411,17 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             if (__pyx_v_met_constraints) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1626
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1625
  * 
  *                             if (met_constraints and
  *                                 self.find_fixpoint(f_x_low, f_back_high,             # <<<<<<<<<<<<<<
  *                                             f_links_low, f_links_high, e_links_low, e_links_high,
  *                                             e_low, e_high, &e_x_low, &e_x_high, &f_x_low, &f_x_high,
  */
-              __pyx_t_14 = PyInt_FromLong(__pyx_v_f_back_high); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1626; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_14 = PyInt_FromLong(__pyx_v_f_back_high); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1625; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_14);
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1630
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1629
  *                                             e_low, e_high, &e_x_low, &e_x_high, &f_x_low, &f_x_high,
  *                                             f_sent_len, e_sent_len,
  *                                             self.train_max_initial_size, self.train_max_initial_size,             # <<<<<<<<<<<<<<
@@ -53379,7 +53431,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
               if (((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->find_fixpoint(__pyx_v_self, __pyx_v_f_x_low, __pyx_t_14, __pyx_v_f_links_low, __pyx_v_f_links_high, __pyx_v_e_links_low, __pyx_v_e_links_high, __pyx_v_e_low, __pyx_v_e_high, (&__pyx_v_e_x_low), (&__pyx_v_e_x_high), (&__pyx_v_f_x_low), (&__pyx_v_f_x_high), __pyx_v_f_sent_len, __pyx_v_e_sent_len, __pyx_v_self->train_max_initial_size, __pyx_v_self->train_max_initial_size, 1, 1, 1, 1, 0, 1, 0)) {
                 __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1632
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1631
  *                                             self.train_max_initial_size, self.train_max_initial_size,
  *                                             1, 1, 1, 1, 0, 1, 0) and
  *                                 ((not self.tight_phrases) or f_links_low[f_x_low] != -1) and             # <<<<<<<<<<<<<<
@@ -53395,17 +53447,17 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 }
                 if (__pyx_t_7) {
 
-                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1633
+                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1632
  *                                             1, 1, 1, 1, 0, 1, 0) and
  *                                 ((not self.tight_phrases) or f_links_low[f_x_low] != -1) and
  *                                 self.find_fixpoint(f_x_low, f_low,    # check integrity of new subphrase             # <<<<<<<<<<<<<<
  *                                             f_links_low, f_links_high, e_links_low, e_links_high,
  *                                             -1, -1, e_gap_low, e_gap_high, f_gap_low, f_gap_high,
  */
-                  __pyx_t_1 = PyInt_FromLong(__pyx_v_f_low); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1633; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_1 = PyInt_FromLong(__pyx_v_f_low); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1632; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   __Pyx_GOTREF(__pyx_t_1);
 
-                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1637
+                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1636
  *                                             -1, -1, e_gap_low, e_gap_high, f_gap_low, f_gap_high,
  *                                             f_sent_len, e_sent_len,
  *                                             self.train_max_initial_size, self.train_max_initial_size,             # <<<<<<<<<<<<<<
@@ -53428,7 +53480,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             }
             if (__pyx_t_9) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1639
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1638
  *                                             self.train_max_initial_size, self.train_max_initial_size,
  *                                             0, 0, 0, 0, 0, 0, 0)):
  *                                 fphr_arr._clear()             # <<<<<<<<<<<<<<
@@ -53437,7 +53489,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
               ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_fphr_arr->__pyx_vtab)->_clear(__pyx_v_fphr_arr);
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1640
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1639
  *                                             0, 0, 0, 0, 0, 0, 0)):
  *                                 fphr_arr._clear()
  *                                 i = 1             # <<<<<<<<<<<<<<
@@ -53446,35 +53498,35 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
               __pyx_v_i = 1;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1641
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1640
  *                                 fphr_arr._clear()
  *                                 i = 1
  *                                 self.findexes.reset()             # <<<<<<<<<<<<<<
  *                                 self.findexes.append(sym_setindex(self.category, i))
  *                                 fphr_arr._append(sym_setindex(self.category, i))
  */
-              __pyx_t_14 = PyObject_GetAttr(((PyObject *)__pyx_v_self->findexes), __pyx_n_s__reset); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1641; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_14 = PyObject_GetAttr(((PyObject *)__pyx_v_self->findexes), __pyx_n_s__reset); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1640; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_14);
-              __pyx_t_1 = PyObject_Call(__pyx_t_14, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1641; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_1 = PyObject_Call(__pyx_t_14, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1640; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_1);
               __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
               __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1642
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1641
  *                                 i = 1
  *                                 self.findexes.reset()
  *                                 self.findexes.append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
  *                                 fphr_arr._append(sym_setindex(self.category, i))
  *                                 i = i+1
  */
-              __pyx_t_1 = PyInt_FromLong(__pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1642; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_1 = PyInt_FromLong(__pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1641; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_1);
-              __pyx_t_14 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes), __pyx_t_1); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1642; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_14 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes), __pyx_t_1); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1641; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_14);
               __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
               __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1643
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1642
  *                                 self.findexes.reset()
  *                                 self.findexes.append(sym_setindex(self.category, i))
  *                                 fphr_arr._append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
@@ -53483,7 +53535,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
               ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_fphr_arr->__pyx_vtab)->_append(__pyx_v_fphr_arr, __pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i));
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1644
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1643
  *                                 self.findexes.append(sym_setindex(self.category, i))
  *                                 fphr_arr._append(sym_setindex(self.category, i))
  *                                 i = i+1             # <<<<<<<<<<<<<<
@@ -53492,27 +53544,27 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
               __pyx_v_i = (__pyx_v_i + 1);
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1645
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1644
  *                                 fphr_arr._append(sym_setindex(self.category, i))
  *                                 i = i+1
  *                                 self.findexes.extend(self.findexes1)             # <<<<<<<<<<<<<<
  *                                 for j from 0 <= j < phrase.n:
  *                                     if sym_isvar(phrase.syms[j]):
  */
-              __pyx_t_14 = PyObject_GetAttr(((PyObject *)__pyx_v_self->findexes), __pyx_n_s__extend); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1645; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_14 = PyObject_GetAttr(((PyObject *)__pyx_v_self->findexes), __pyx_n_s__extend); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1644; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_14);
-              __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1645; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1644; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_1);
               __Pyx_INCREF(((PyObject *)__pyx_v_self->findexes1));
               PyTuple_SET_ITEM(__pyx_t_1, 0, ((PyObject *)__pyx_v_self->findexes1));
               __Pyx_GIVEREF(((PyObject *)__pyx_v_self->findexes1));
-              __pyx_t_15 = PyObject_Call(__pyx_t_14, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1645; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_15 = PyObject_Call(__pyx_t_14, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1644; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_15);
               __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
               __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
               __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1646
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1645
  *                                 i = i+1
  *                                 self.findexes.extend(self.findexes1)
  *                                 for j from 0 <= j < phrase.n:             # <<<<<<<<<<<<<<
@@ -53522,7 +53574,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
               __pyx_t_3 = __pyx_v_phrase->n;
               for (__pyx_v_j = 0; __pyx_v_j < __pyx_t_3; __pyx_v_j++) {
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1647
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1646
  *                                 self.findexes.extend(self.findexes1)
  *                                 for j from 0 <= j < phrase.n:
  *                                     if sym_isvar(phrase.syms[j]):             # <<<<<<<<<<<<<<
@@ -53532,7 +53584,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 __pyx_t_4 = __pyx_f_3_sa_sym_isvar((__pyx_v_phrase->syms[__pyx_v_j]));
                 if (__pyx_t_4) {
 
-                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1648
+                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1647
  *                                 for j from 0 <= j < phrase.n:
  *                                     if sym_isvar(phrase.syms[j]):
  *                                         fphr_arr._append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
@@ -53541,7 +53593,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
                   ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_fphr_arr->__pyx_vtab)->_append(__pyx_v_fphr_arr, __pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i));
 
-                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1649
+                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1648
  *                                     if sym_isvar(phrase.syms[j]):
  *                                         fphr_arr._append(sym_setindex(self.category, i))
  *                                         i = i + 1             # <<<<<<<<<<<<<<
@@ -53553,7 +53605,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 }
                 /*else*/ {
 
-                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1651
+                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1650
  *                                         i = i + 1
  *                                     else:
  *                                         fphr_arr._append(phrase.syms[j])             # <<<<<<<<<<<<<<
@@ -53565,7 +53617,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 __pyx_L72:;
               }
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1652
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1651
  *                                     else:
  *                                         fphr_arr._append(phrase.syms[j])
  *                                 if f_back_high > f_high:             # <<<<<<<<<<<<<<
@@ -53575,7 +53627,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
               __pyx_t_9 = (__pyx_v_f_back_high > __pyx_v_f_high);
               if (__pyx_t_9) {
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1653
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1652
  *                                         fphr_arr._append(phrase.syms[j])
  *                                 if f_back_high > f_high:
  *                                     fphr_arr._append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
@@ -53584,16 +53636,16 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
                 ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_fphr_arr->__pyx_vtab)->_append(__pyx_v_fphr_arr, __pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i));
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1654
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1653
  *                                 if f_back_high > f_high:
  *                                     fphr_arr._append(sym_setindex(self.category, i))
  *                                     self.findexes.append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
  *                                 fphr = Phrase(fphr_arr)
  *                                 phrase_list = self.extract_phrases(e_x_low, e_x_high, e_gap_low, e_gap_high, e_links_low, num_gaps+1,
  */
-                __pyx_t_15 = PyInt_FromLong(__pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i)); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1654; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_15 = PyInt_FromLong(__pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i)); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1653; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_15);
-                __pyx_t_1 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes), __pyx_t_15); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1654; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_1 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes), __pyx_t_15); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1653; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_1);
                 __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
                 __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
@@ -53601,67 +53653,67 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
               }
               __pyx_L73:;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1655
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1654
  *                                     fphr_arr._append(sym_setindex(self.category, i))
  *                                     self.findexes.append(sym_setindex(self.category, i))
  *                                 fphr = Phrase(fphr_arr)             # <<<<<<<<<<<<<<
  *                                 phrase_list = self.extract_phrases(e_x_low, e_x_high, e_gap_low, e_gap_high, e_links_low, num_gaps+1,
  *                                                     f_x_low, f_x_high, f_gap_low, f_gap_high, f_links_low, matching.sent_id,
  */
-              __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1655; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1654; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_1);
               __Pyx_INCREF(((PyObject *)__pyx_v_fphr_arr));
               PyTuple_SET_ITEM(__pyx_t_1, 0, ((PyObject *)__pyx_v_fphr_arr));
               __Pyx_GIVEREF(((PyObject *)__pyx_v_fphr_arr));
-              __pyx_t_15 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_Phrase)), ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1655; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_15 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_Phrase)), ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1654; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_15);
               __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
               __Pyx_DECREF(((PyObject *)__pyx_v_fphr));
               __pyx_v_fphr = ((struct __pyx_obj_3_sa_Phrase *)__pyx_t_15);
               __pyx_t_15 = 0;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1658
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1657
  *                                 phrase_list = self.extract_phrases(e_x_low, e_x_high, e_gap_low, e_gap_high, e_links_low, num_gaps+1,
  *                                                     f_x_low, f_x_high, f_gap_low, f_gap_high, f_links_low, matching.sent_id,
  *                                                     e_sent_len, e_sent_start)             # <<<<<<<<<<<<<<
  *                                 if len(phrase_list) > 0:
  *                                     pair_count = 1.0 / len(phrase_list)
  */
-              __pyx_t_15 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->extract_phrases(__pyx_v_self, __pyx_v_e_x_low, __pyx_v_e_x_high, __pyx_v_e_gap_low, __pyx_v_e_gap_high, __pyx_v_e_links_low, (__pyx_v_num_gaps + 1), __pyx_v_f_x_low, __pyx_v_f_x_high, __pyx_v_f_gap_low, __pyx_v_f_gap_high, __pyx_v_f_links_low, __pyx_v_matching->sent_id, __pyx_v_e_sent_len, __pyx_v_e_sent_start); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1656; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_15 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->extract_phrases(__pyx_v_self, __pyx_v_e_x_low, __pyx_v_e_x_high, __pyx_v_e_gap_low, __pyx_v_e_gap_high, __pyx_v_e_links_low, (__pyx_v_num_gaps + 1), __pyx_v_f_x_low, __pyx_v_f_x_high, __pyx_v_f_gap_low, __pyx_v_f_gap_high, __pyx_v_f_links_low, __pyx_v_matching->sent_id, __pyx_v_e_sent_len, __pyx_v_e_sent_start); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1655; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_15);
               __Pyx_XDECREF(__pyx_v_phrase_list);
               __pyx_v_phrase_list = __pyx_t_15;
               __pyx_t_15 = 0;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1659
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1658
  *                                                     f_x_low, f_x_high, f_gap_low, f_gap_high, f_links_low, matching.sent_id,
  *                                                     e_sent_len, e_sent_start)
  *                                 if len(phrase_list) > 0:             # <<<<<<<<<<<<<<
  *                                     pair_count = 1.0 / len(phrase_list)
  *                                 else:
  */
-              __pyx_t_13 = PyObject_Length(__pyx_v_phrase_list); if (unlikely(__pyx_t_13 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1659; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_13 = PyObject_Length(__pyx_v_phrase_list); if (unlikely(__pyx_t_13 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1658; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __pyx_t_9 = (__pyx_t_13 > 0);
               if (__pyx_t_9) {
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1660
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1659
  *                                                     e_sent_len, e_sent_start)
  *                                 if len(phrase_list) > 0:
  *                                     pair_count = 1.0 / len(phrase_list)             # <<<<<<<<<<<<<<
  *                                 else:
  *                                     pair_count = 0
  */
-                __pyx_t_13 = PyObject_Length(__pyx_v_phrase_list); if (unlikely(__pyx_t_13 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1660; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_13 = PyObject_Length(__pyx_v_phrase_list); if (unlikely(__pyx_t_13 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1659; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 if (unlikely(__pyx_t_13 == 0)) {
                   PyErr_Format(PyExc_ZeroDivisionError, "float division");
-                  {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1660; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1659; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 }
                 __pyx_v_pair_count = (1.0 / __pyx_t_13);
                 goto __pyx_L74;
               }
               /*else*/ {
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1662
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1661
  *                                     pair_count = 1.0 / len(phrase_list)
  *                                 else:
  *                                     pair_count = 0             # <<<<<<<<<<<<<<
@@ -53672,7 +53724,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
               }
               __pyx_L74:;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1663
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1662
  *                                 else:
  *                                     pair_count = 0
  *                                 for phrase2,eindexes in phrase_list:             # <<<<<<<<<<<<<<
@@ -53683,7 +53735,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 __pyx_t_15 = __pyx_v_phrase_list; __Pyx_INCREF(__pyx_t_15); __pyx_t_13 = 0;
                 __pyx_t_16 = NULL;
               } else {
-                __pyx_t_13 = -1; __pyx_t_15 = PyObject_GetIter(__pyx_v_phrase_list); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_13 = -1; __pyx_t_15 = PyObject_GetIter(__pyx_v_phrase_list); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1662; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_15);
                 __pyx_t_16 = Py_TYPE(__pyx_t_15)->tp_iternext;
               }
@@ -53691,23 +53743,23 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 if (!__pyx_t_16 && PyList_CheckExact(__pyx_t_15)) {
                   if (__pyx_t_13 >= PyList_GET_SIZE(__pyx_t_15)) break;
                   #if CYTHON_COMPILING_IN_CPYTHON
-                  __pyx_t_1 = PyList_GET_ITEM(__pyx_t_15, __pyx_t_13); __Pyx_INCREF(__pyx_t_1); __pyx_t_13++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_1 = PyList_GET_ITEM(__pyx_t_15, __pyx_t_13); __Pyx_INCREF(__pyx_t_1); __pyx_t_13++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1662; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   #else
-                  __pyx_t_1 = PySequence_ITEM(__pyx_t_15, __pyx_t_13); __pyx_t_13++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_1 = PySequence_ITEM(__pyx_t_15, __pyx_t_13); __pyx_t_13++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1662; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   #endif
                 } else if (!__pyx_t_16 && PyTuple_CheckExact(__pyx_t_15)) {
                   if (__pyx_t_13 >= PyTuple_GET_SIZE(__pyx_t_15)) break;
                   #if CYTHON_COMPILING_IN_CPYTHON
-                  __pyx_t_1 = PyTuple_GET_ITEM(__pyx_t_15, __pyx_t_13); __Pyx_INCREF(__pyx_t_1); __pyx_t_13++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_1 = PyTuple_GET_ITEM(__pyx_t_15, __pyx_t_13); __Pyx_INCREF(__pyx_t_1); __pyx_t_13++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1662; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   #else
-                  __pyx_t_1 = PySequence_ITEM(__pyx_t_15, __pyx_t_13); __pyx_t_13++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_1 = PySequence_ITEM(__pyx_t_15, __pyx_t_13); __pyx_t_13++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1662; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   #endif
                 } else {
                   __pyx_t_1 = __pyx_t_16(__pyx_t_15);
                   if (unlikely(!__pyx_t_1)) {
                     if (PyErr_Occurred()) {
                       if (likely(PyErr_ExceptionMatches(PyExc_StopIteration))) PyErr_Clear();
-                      else {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                      else {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1662; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                     }
                     break;
                   }
@@ -53723,7 +53775,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                   if (unlikely(size != 2)) {
                     if (size > 2) __Pyx_RaiseTooManyValuesError(2);
                     else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size);
-                    {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                    {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1662; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   }
                   #if CYTHON_COMPILING_IN_CPYTHON
                   if (likely(PyTuple_CheckExact(sequence))) {
@@ -53736,14 +53788,14 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                   __Pyx_INCREF(__pyx_t_14);
                   __Pyx_INCREF(__pyx_t_2);
                   #else
-                  __pyx_t_14 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-                  __pyx_t_2 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_14 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1662; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_2 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1662; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   #endif
                   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
                 } else
                 {
                   Py_ssize_t index = -1;
-                  __pyx_t_10 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_10 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1662; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   __Pyx_GOTREF(__pyx_t_10);
                   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
                   __pyx_t_17 = Py_TYPE(__pyx_t_10)->tp_iternext;
@@ -53751,7 +53803,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                   __Pyx_GOTREF(__pyx_t_14);
                   index = 1; __pyx_t_2 = __pyx_t_17(__pyx_t_10); if (unlikely(!__pyx_t_2)) goto __pyx_L77_unpacking_failed;
                   __Pyx_GOTREF(__pyx_t_2);
-                  if (__Pyx_IternextUnpackEndCheck(__pyx_t_17(__pyx_t_10), 2) < 0) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  if (__Pyx_IternextUnpackEndCheck(__pyx_t_17(__pyx_t_10), 2) < 0) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1662; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   __pyx_t_17 = NULL;
                   __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
                   goto __pyx_L78_unpacking_done;
@@ -53759,7 +53811,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                   __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
                   __pyx_t_17 = NULL;
                   if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index);
-                  {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1662; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   __pyx_L78_unpacking_done:;
                 }
                 __Pyx_XDECREF(__pyx_v_phrase2);
@@ -53769,7 +53821,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 __pyx_v_eindexes = __pyx_t_2;
                 __pyx_t_2 = 0;
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1664
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1663
  *                                     pair_count = 0
  *                                 for phrase2,eindexes in phrase_list:
  *                                     als2 = self.create_alignments(sent_links,num_links,self.findexes,eindexes)             # <<<<<<<<<<<<<<
@@ -53778,31 +53830,31 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
                 __pyx_t_1 = ((PyObject *)__pyx_v_self->findexes);
                 __Pyx_INCREF(__pyx_t_1);
-                __pyx_t_2 = ((PyObject *)((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->create_alignments(__pyx_v_self, __pyx_v_sent_links, __pyx_v_num_links, __pyx_t_1, __pyx_v_eindexes)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1664; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_2 = ((PyObject *)((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->create_alignments(__pyx_v_self, __pyx_v_sent_links, __pyx_v_num_links, __pyx_t_1, __pyx_v_eindexes)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_2);
                 __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
                 __Pyx_XDECREF(((PyObject *)__pyx_v_als2));
                 __pyx_v_als2 = ((struct __pyx_obj_3_sa_IntList *)__pyx_t_2);
                 __pyx_t_2 = 0;
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1665
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1664
  *                                 for phrase2,eindexes in phrase_list:
  *                                     als2 = self.create_alignments(sent_links,num_links,self.findexes,eindexes)
  *                                     extracts.append((fphr, phrase2, pair_count, tuple(als2)))             # <<<<<<<<<<<<<<
  * 
  *                         if (f_back_high == f_high and
  */
-                __pyx_t_2 = PyFloat_FromDouble(__pyx_v_pair_count); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1665; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_2 = PyFloat_FromDouble(__pyx_v_pair_count); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1664; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_2);
-                __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1665; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1664; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_1);
                 __Pyx_INCREF(((PyObject *)__pyx_v_als2));
                 PyTuple_SET_ITEM(__pyx_t_1, 0, ((PyObject *)__pyx_v_als2));
                 __Pyx_GIVEREF(((PyObject *)__pyx_v_als2));
-                __pyx_t_14 = PyObject_Call(((PyObject *)((PyObject*)(&PyTuple_Type))), ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1665; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_14 = PyObject_Call(((PyObject *)((PyObject*)(&PyTuple_Type))), ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1664; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_14);
                 __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
-                __pyx_t_1 = PyTuple_New(4); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1665; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_1 = PyTuple_New(4); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1664; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_1);
                 __Pyx_INCREF(((PyObject *)__pyx_v_fphr));
                 PyTuple_SET_ITEM(__pyx_t_1, 0, ((PyObject *)__pyx_v_fphr));
@@ -53816,7 +53868,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 __Pyx_GIVEREF(__pyx_t_14);
                 __pyx_t_2 = 0;
                 __pyx_t_14 = 0;
-                __pyx_t_14 = __Pyx_PyObject_Append(__pyx_v_extracts, ((PyObject *)__pyx_t_1)); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1665; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_14 = __Pyx_PyObject_Append(__pyx_v_extracts, ((PyObject *)__pyx_t_1)); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1664; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_14);
                 __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
                 __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
@@ -53829,7 +53881,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
           }
           __pyx_L64:;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1667
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1666
  *                                     extracts.append((fphr, phrase2, pair_count, tuple(als2)))
  * 
  *                         if (f_back_high == f_high and             # <<<<<<<<<<<<<<
@@ -53839,7 +53891,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
           __pyx_t_9 = (__pyx_v_f_back_high == __pyx_v_f_high);
           if (__pyx_t_9) {
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1668
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1667
  * 
  *                         if (f_back_high == f_high and
  *                             f_sent_len - f_high >= self.train_min_gap_size and             # <<<<<<<<<<<<<<
@@ -53849,7 +53901,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             __pyx_t_7 = ((__pyx_v_f_sent_len - __pyx_v_f_high) >= __pyx_v_self->train_min_gap_size);
             if (__pyx_t_7) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1669
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1668
  *                         if (f_back_high == f_high and
  *                             f_sent_len - f_high >= self.train_min_gap_size and
  *                             ((not self.tight_phrases) or (f_links_low[f_high] != -1 and f_links_low[f_back_low] != -1))):             # <<<<<<<<<<<<<<
@@ -53879,7 +53931,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
           }
           if (__pyx_t_7) {
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1670
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1669
  *                             f_sent_len - f_high >= self.train_min_gap_size and
  *                             ((not self.tight_phrases) or (f_links_low[f_high] != -1 and f_links_low[f_back_low] != -1))):
  *                             f_x_high = f_high+self.train_min_gap_size             # <<<<<<<<<<<<<<
@@ -53888,7 +53940,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             __pyx_v_f_x_high = (__pyx_v_f_high + __pyx_v_self->train_min_gap_size);
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1671
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1670
  *                             ((not self.tight_phrases) or (f_links_low[f_high] != -1 and f_links_low[f_back_low] != -1))):
  *                             f_x_high = f_high+self.train_min_gap_size
  *                             met_constraints = 1             # <<<<<<<<<<<<<<
@@ -53897,7 +53949,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             __pyx_v_met_constraints = 1;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1672
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1671
  *                             f_x_high = f_high+self.train_min_gap_size
  *                             met_constraints = 1
  *                             if self.tight_phrases:             # <<<<<<<<<<<<<<
@@ -53906,7 +53958,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             if (__pyx_v_self->tight_phrases) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1673
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1672
  *                             met_constraints = 1
  *                             if self.tight_phrases:
  *                                 while f_x_high <= f_sent_len and f_links_low[f_x_high-1] == -1:             # <<<<<<<<<<<<<<
@@ -53923,7 +53975,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 }
                 if (!__pyx_t_18) break;
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1674
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1673
  *                             if self.tight_phrases:
  *                                 while f_x_high <= f_sent_len and f_links_low[f_x_high-1] == -1:
  *                                     f_x_high = f_x_high + 1             # <<<<<<<<<<<<<<
@@ -53936,7 +53988,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             }
             __pyx_L80:;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1675
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1674
  *                                 while f_x_high <= f_sent_len and f_links_low[f_x_high-1] == -1:
  *                                     f_x_high = f_x_high + 1
  *                             if f_x_high > f_sent_len or f_x_high - f_back_low > self.train_max_initial_size:             # <<<<<<<<<<<<<<
@@ -53952,7 +54004,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             }
             if (__pyx_t_9) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1676
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1675
  *                                     f_x_high = f_x_high + 1
  *                             if f_x_high > f_sent_len or f_x_high - f_back_low > self.train_max_initial_size:
  *                                 met_constraints = 0             # <<<<<<<<<<<<<<
@@ -53964,7 +54016,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             }
             __pyx_L83:;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1678
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1677
  *                                 met_constraints = 0
  * 
  *                             if (met_constraints and             # <<<<<<<<<<<<<<
@@ -53973,17 +54025,17 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             if (__pyx_v_met_constraints) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1679
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1678
  * 
  *                             if (met_constraints and
  *                                 self.find_fixpoint(f_back_low, f_x_high,             # <<<<<<<<<<<<<<
  *                                             f_links_low, f_links_high, e_links_low, e_links_high,
  *                                             e_low, e_high, &e_x_low, &e_x_high, &f_x_low, &f_x_high,
  */
-              __pyx_t_15 = PyInt_FromLong(__pyx_v_f_x_high); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1679; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_15 = PyInt_FromLong(__pyx_v_f_x_high); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1678; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_15);
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1683
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1682
  *                                             e_low, e_high, &e_x_low, &e_x_high, &f_x_low, &f_x_high,
  *                                             f_sent_len, e_sent_len,
  *                                             self.train_max_initial_size, self.train_max_initial_size,             # <<<<<<<<<<<<<<
@@ -53993,7 +54045,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
               if (((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->find_fixpoint(__pyx_v_self, __pyx_v_f_back_low, __pyx_t_15, __pyx_v_f_links_low, __pyx_v_f_links_high, __pyx_v_e_links_low, __pyx_v_e_links_high, __pyx_v_e_low, __pyx_v_e_high, (&__pyx_v_e_x_low), (&__pyx_v_e_x_high), (&__pyx_v_f_x_low), (&__pyx_v_f_x_high), __pyx_v_f_sent_len, __pyx_v_e_sent_len, __pyx_v_self->train_max_initial_size, __pyx_v_self->train_max_initial_size, 1, 1, 1, 0, 1, 1, 0)) {
                 __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1685
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1684
  *                                             self.train_max_initial_size, self.train_max_initial_size,
  *                                             1, 1, 1, 0, 1, 1, 0) and
  *                                 ((not self.tight_phrases) or f_links_low[f_x_high-1] != -1) and             # <<<<<<<<<<<<<<
@@ -54009,17 +54061,17 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 }
                 if (__pyx_t_7) {
 
-                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1686
+                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1685
  *                                             1, 1, 1, 0, 1, 1, 0) and
  *                                 ((not self.tight_phrases) or f_links_low[f_x_high-1] != -1) and
  *                                 self.find_fixpoint(f_high, f_x_high,             # <<<<<<<<<<<<<<
  *                                             f_links_low, f_links_high, e_links_low, e_links_high,
  *                                             -1, -1, e_gap_low+gap_start+num_gaps, e_gap_high+gap_start+num_gaps,
  */
-                  __pyx_t_14 = PyInt_FromLong(__pyx_v_f_x_high); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1686; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_14 = PyInt_FromLong(__pyx_v_f_x_high); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1685; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   __Pyx_GOTREF(__pyx_t_14);
 
-                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1691
+                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1690
  *                                             f_gap_low+gap_start+num_gaps, f_gap_high+gap_start+num_gaps,
  *                                             f_sent_len, e_sent_len,
  *                                             self.train_max_initial_size, self.train_max_initial_size,             # <<<<<<<<<<<<<<
@@ -54042,7 +54094,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             }
             if (__pyx_t_9) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1693
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1692
  *                                             self.train_max_initial_size, self.train_max_initial_size,
  *                                             0, 0, 0, 0, 0, 0, 0)):
  *                                 fphr_arr._clear()             # <<<<<<<<<<<<<<
@@ -54051,7 +54103,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
               ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_fphr_arr->__pyx_vtab)->_clear(__pyx_v_fphr_arr);
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1694
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1693
  *                                             0, 0, 0, 0, 0, 0, 0)):
  *                                 fphr_arr._clear()
  *                                 i = 1             # <<<<<<<<<<<<<<
@@ -54060,21 +54112,21 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
               __pyx_v_i = 1;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1695
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1694
  *                                 fphr_arr._clear()
  *                                 i = 1
  *                                 self.findexes.reset()             # <<<<<<<<<<<<<<
  *                                 if f_back_low < f_low:
  *                                     fphr_arr._append(sym_setindex(self.category, i))
  */
-              __pyx_t_15 = PyObject_GetAttr(((PyObject *)__pyx_v_self->findexes), __pyx_n_s__reset); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1695; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_15 = PyObject_GetAttr(((PyObject *)__pyx_v_self->findexes), __pyx_n_s__reset); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1694; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_15);
-              __pyx_t_14 = PyObject_Call(__pyx_t_15, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1695; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_14 = PyObject_Call(__pyx_t_15, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1694; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_14);
               __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
               __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1696
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1695
  *                                 i = 1
  *                                 self.findexes.reset()
  *                                 if f_back_low < f_low:             # <<<<<<<<<<<<<<
@@ -54084,7 +54136,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
               __pyx_t_9 = (__pyx_v_f_back_low < __pyx_v_f_low);
               if (__pyx_t_9) {
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1697
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1696
  *                                 self.findexes.reset()
  *                                 if f_back_low < f_low:
  *                                     fphr_arr._append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
@@ -54093,7 +54145,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
                 ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_fphr_arr->__pyx_vtab)->_append(__pyx_v_fphr_arr, __pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i));
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1698
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1697
  *                                 if f_back_low < f_low:
  *                                     fphr_arr._append(sym_setindex(self.category, i))
  *                                     i = i+1             # <<<<<<<<<<<<<<
@@ -54102,16 +54154,16 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
                 __pyx_v_i = (__pyx_v_i + 1);
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1699
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1698
  *                                     fphr_arr._append(sym_setindex(self.category, i))
  *                                     i = i+1
  *                                     self.findexes.append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
  *                                 self.findexes.extend(self.findexes1)
  *                                 for j from 0 <= j < phrase.n:
  */
-                __pyx_t_14 = PyInt_FromLong(__pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i)); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1699; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_14 = PyInt_FromLong(__pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i)); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1698; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_14);
-                __pyx_t_15 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes), __pyx_t_14); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1699; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_15 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes), __pyx_t_14); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1698; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_15);
                 __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
                 __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
@@ -54119,27 +54171,27 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
               }
               __pyx_L85:;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1700
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1699
  *                                     i = i+1
  *                                     self.findexes.append(sym_setindex(self.category, i))
  *                                 self.findexes.extend(self.findexes1)             # <<<<<<<<<<<<<<
  *                                 for j from 0 <= j < phrase.n:
  *                                     if sym_isvar(phrase.syms[j]):
  */
-              __pyx_t_15 = PyObject_GetAttr(((PyObject *)__pyx_v_self->findexes), __pyx_n_s__extend); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1700; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_15 = PyObject_GetAttr(((PyObject *)__pyx_v_self->findexes), __pyx_n_s__extend); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1699; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_15);
-              __pyx_t_14 = PyTuple_New(1); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1700; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_14 = PyTuple_New(1); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1699; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_14);
               __Pyx_INCREF(((PyObject *)__pyx_v_self->findexes1));
               PyTuple_SET_ITEM(__pyx_t_14, 0, ((PyObject *)__pyx_v_self->findexes1));
               __Pyx_GIVEREF(((PyObject *)__pyx_v_self->findexes1));
-              __pyx_t_1 = PyObject_Call(__pyx_t_15, ((PyObject *)__pyx_t_14), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1700; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_1 = PyObject_Call(__pyx_t_15, ((PyObject *)__pyx_t_14), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1699; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_1);
               __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
               __Pyx_DECREF(((PyObject *)__pyx_t_14)); __pyx_t_14 = 0;
               __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1701
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1700
  *                                     self.findexes.append(sym_setindex(self.category, i))
  *                                 self.findexes.extend(self.findexes1)
  *                                 for j from 0 <= j < phrase.n:             # <<<<<<<<<<<<<<
@@ -54149,7 +54201,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
               __pyx_t_3 = __pyx_v_phrase->n;
               for (__pyx_v_j = 0; __pyx_v_j < __pyx_t_3; __pyx_v_j++) {
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1702
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1701
  *                                 self.findexes.extend(self.findexes1)
  *                                 for j from 0 <= j < phrase.n:
  *                                     if sym_isvar(phrase.syms[j]):             # <<<<<<<<<<<<<<
@@ -54159,7 +54211,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 __pyx_t_4 = __pyx_f_3_sa_sym_isvar((__pyx_v_phrase->syms[__pyx_v_j]));
                 if (__pyx_t_4) {
 
-                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1703
+                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1702
  *                                 for j from 0 <= j < phrase.n:
  *                                     if sym_isvar(phrase.syms[j]):
  *                                         fphr_arr._append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
@@ -54168,7 +54220,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
                   ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_fphr_arr->__pyx_vtab)->_append(__pyx_v_fphr_arr, __pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i));
 
-                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1704
+                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1703
  *                                     if sym_isvar(phrase.syms[j]):
  *                                         fphr_arr._append(sym_setindex(self.category, i))
  *                                         i = i + 1             # <<<<<<<<<<<<<<
@@ -54180,7 +54232,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 }
                 /*else*/ {
 
-                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1706
+                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1705
  *                                         i = i + 1
  *                                     else:
  *                                         fphr_arr._append(phrase.syms[j])             # <<<<<<<<<<<<<<
@@ -54192,7 +54244,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 __pyx_L88:;
               }
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1707
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1706
  *                                     else:
  *                                         fphr_arr._append(phrase.syms[j])
  *                                 fphr_arr._append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
@@ -54201,81 +54253,81 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
               ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_fphr_arr->__pyx_vtab)->_append(__pyx_v_fphr_arr, __pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i));
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1708
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1707
  *                                         fphr_arr._append(phrase.syms[j])
  *                                 fphr_arr._append(sym_setindex(self.category, i))
  *                                 self.findexes.append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
  *                                 fphr = Phrase(fphr_arr)
  *                                 phrase_list = self.extract_phrases(e_x_low, e_x_high, e_gap_low+gap_start, e_gap_high+gap_start, e_links_low, num_gaps+1,
  */
-              __pyx_t_1 = PyInt_FromLong(__pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1708; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_1 = PyInt_FromLong(__pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1707; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_1);
-              __pyx_t_14 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes), __pyx_t_1); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1708; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_14 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes), __pyx_t_1); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1707; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_14);
               __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
               __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1709
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1708
  *                                 fphr_arr._append(sym_setindex(self.category, i))
  *                                 self.findexes.append(sym_setindex(self.category, i))
  *                                 fphr = Phrase(fphr_arr)             # <<<<<<<<<<<<<<
  *                                 phrase_list = self.extract_phrases(e_x_low, e_x_high, e_gap_low+gap_start, e_gap_high+gap_start, e_links_low, num_gaps+1,
  *                                                     f_x_low, f_x_high, f_gap_low+gap_start, f_gap_high+gap_start, f_links_low,
  */
-              __pyx_t_14 = PyTuple_New(1); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1709; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_14 = PyTuple_New(1); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1708; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_14);
               __Pyx_INCREF(((PyObject *)__pyx_v_fphr_arr));
               PyTuple_SET_ITEM(__pyx_t_14, 0, ((PyObject *)__pyx_v_fphr_arr));
               __Pyx_GIVEREF(((PyObject *)__pyx_v_fphr_arr));
-              __pyx_t_1 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_Phrase)), ((PyObject *)__pyx_t_14), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1709; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_1 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_Phrase)), ((PyObject *)__pyx_t_14), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1708; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_1);
               __Pyx_DECREF(((PyObject *)__pyx_t_14)); __pyx_t_14 = 0;
               __Pyx_DECREF(((PyObject *)__pyx_v_fphr));
               __pyx_v_fphr = ((struct __pyx_obj_3_sa_Phrase *)__pyx_t_1);
               __pyx_t_1 = 0;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1712
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1711
  *                                 phrase_list = self.extract_phrases(e_x_low, e_x_high, e_gap_low+gap_start, e_gap_high+gap_start, e_links_low, num_gaps+1,
  *                                                     f_x_low, f_x_high, f_gap_low+gap_start, f_gap_high+gap_start, f_links_low,
  *                                                     matching.sent_id, e_sent_len, e_sent_start)             # <<<<<<<<<<<<<<
  *                                 if len(phrase_list) > 0:
  *                                     pair_count = 1.0 / len(phrase_list)
  */
-              __pyx_t_1 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->extract_phrases(__pyx_v_self, __pyx_v_e_x_low, __pyx_v_e_x_high, (__pyx_v_e_gap_low + __pyx_v_gap_start), (__pyx_v_e_gap_high + __pyx_v_gap_start), __pyx_v_e_links_low, (__pyx_v_num_gaps + 1), __pyx_v_f_x_low, __pyx_v_f_x_high, (__pyx_v_f_gap_low + __pyx_v_gap_start), (__pyx_v_f_gap_high + __pyx_v_gap_start), __pyx_v_f_links_low, __pyx_v_matching->sent_id, __pyx_v_e_sent_len, __pyx_v_e_sent_start); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1710; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_1 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->extract_phrases(__pyx_v_self, __pyx_v_e_x_low, __pyx_v_e_x_high, (__pyx_v_e_gap_low + __pyx_v_gap_start), (__pyx_v_e_gap_high + __pyx_v_gap_start), __pyx_v_e_links_low, (__pyx_v_num_gaps + 1), __pyx_v_f_x_low, __pyx_v_f_x_high, (__pyx_v_f_gap_low + __pyx_v_gap_start), (__pyx_v_f_gap_high + __pyx_v_gap_start), __pyx_v_f_links_low, __pyx_v_matching->sent_id, __pyx_v_e_sent_len, __pyx_v_e_sent_start); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1709; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_1);
               __Pyx_XDECREF(__pyx_v_phrase_list);
               __pyx_v_phrase_list = __pyx_t_1;
               __pyx_t_1 = 0;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1713
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1712
  *                                                     f_x_low, f_x_high, f_gap_low+gap_start, f_gap_high+gap_start, f_links_low,
  *                                                     matching.sent_id, e_sent_len, e_sent_start)
  *                                 if len(phrase_list) > 0:             # <<<<<<<<<<<<<<
  *                                     pair_count = 1.0 / len(phrase_list)
  *                                 else:
  */
-              __pyx_t_13 = PyObject_Length(__pyx_v_phrase_list); if (unlikely(__pyx_t_13 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1713; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_13 = PyObject_Length(__pyx_v_phrase_list); if (unlikely(__pyx_t_13 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1712; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __pyx_t_9 = (__pyx_t_13 > 0);
               if (__pyx_t_9) {
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1714
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1713
  *                                                     matching.sent_id, e_sent_len, e_sent_start)
  *                                 if len(phrase_list) > 0:
  *                                     pair_count = 1.0 / len(phrase_list)             # <<<<<<<<<<<<<<
  *                                 else:
  *                                     pair_count = 0
  */
-                __pyx_t_13 = PyObject_Length(__pyx_v_phrase_list); if (unlikely(__pyx_t_13 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1714; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_13 = PyObject_Length(__pyx_v_phrase_list); if (unlikely(__pyx_t_13 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1713; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 if (unlikely(__pyx_t_13 == 0)) {
                   PyErr_Format(PyExc_ZeroDivisionError, "float division");
-                  {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1714; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1713; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 }
                 __pyx_v_pair_count = (1.0 / __pyx_t_13);
                 goto __pyx_L89;
               }
               /*else*/ {
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1716
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1715
  *                                     pair_count = 1.0 / len(phrase_list)
  *                                 else:
  *                                     pair_count = 0             # <<<<<<<<<<<<<<
@@ -54286,7 +54338,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
               }
               __pyx_L89:;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1717
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1716
  *                                 else:
  *                                     pair_count = 0
  *                                 for phrase2, eindexes in phrase_list:             # <<<<<<<<<<<<<<
@@ -54297,7 +54349,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 __pyx_t_1 = __pyx_v_phrase_list; __Pyx_INCREF(__pyx_t_1); __pyx_t_13 = 0;
                 __pyx_t_16 = NULL;
               } else {
-                __pyx_t_13 = -1; __pyx_t_1 = PyObject_GetIter(__pyx_v_phrase_list); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1717; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_13 = -1; __pyx_t_1 = PyObject_GetIter(__pyx_v_phrase_list); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1716; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_1);
                 __pyx_t_16 = Py_TYPE(__pyx_t_1)->tp_iternext;
               }
@@ -54305,23 +54357,23 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 if (!__pyx_t_16 && PyList_CheckExact(__pyx_t_1)) {
                   if (__pyx_t_13 >= PyList_GET_SIZE(__pyx_t_1)) break;
                   #if CYTHON_COMPILING_IN_CPYTHON
-                  __pyx_t_14 = PyList_GET_ITEM(__pyx_t_1, __pyx_t_13); __Pyx_INCREF(__pyx_t_14); __pyx_t_13++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1717; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_14 = PyList_GET_ITEM(__pyx_t_1, __pyx_t_13); __Pyx_INCREF(__pyx_t_14); __pyx_t_13++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1716; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   #else
-                  __pyx_t_14 = PySequence_ITEM(__pyx_t_1, __pyx_t_13); __pyx_t_13++; if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1717; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_14 = PySequence_ITEM(__pyx_t_1, __pyx_t_13); __pyx_t_13++; if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1716; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   #endif
                 } else if (!__pyx_t_16 && PyTuple_CheckExact(__pyx_t_1)) {
                   if (__pyx_t_13 >= PyTuple_GET_SIZE(__pyx_t_1)) break;
                   #if CYTHON_COMPILING_IN_CPYTHON
-                  __pyx_t_14 = PyTuple_GET_ITEM(__pyx_t_1, __pyx_t_13); __Pyx_INCREF(__pyx_t_14); __pyx_t_13++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1717; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_14 = PyTuple_GET_ITEM(__pyx_t_1, __pyx_t_13); __Pyx_INCREF(__pyx_t_14); __pyx_t_13++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1716; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   #else
-                  __pyx_t_14 = PySequence_ITEM(__pyx_t_1, __pyx_t_13); __pyx_t_13++; if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1717; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_14 = PySequence_ITEM(__pyx_t_1, __pyx_t_13); __pyx_t_13++; if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1716; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   #endif
                 } else {
                   __pyx_t_14 = __pyx_t_16(__pyx_t_1);
                   if (unlikely(!__pyx_t_14)) {
                     if (PyErr_Occurred()) {
                       if (likely(PyErr_ExceptionMatches(PyExc_StopIteration))) PyErr_Clear();
-                      else {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1717; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                      else {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1716; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                     }
                     break;
                   }
@@ -54337,7 +54389,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                   if (unlikely(size != 2)) {
                     if (size > 2) __Pyx_RaiseTooManyValuesError(2);
                     else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size);
-                    {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1717; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                    {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1716; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   }
                   #if CYTHON_COMPILING_IN_CPYTHON
                   if (likely(PyTuple_CheckExact(sequence))) {
@@ -54350,14 +54402,14 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                   __Pyx_INCREF(__pyx_t_15);
                   __Pyx_INCREF(__pyx_t_2);
                   #else
-                  __pyx_t_15 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1717; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-                  __pyx_t_2 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1717; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_15 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1716; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_2 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1716; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   #endif
                   __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
                 } else
                 {
                   Py_ssize_t index = -1;
-                  __pyx_t_10 = PyObject_GetIter(__pyx_t_14); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1717; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_10 = PyObject_GetIter(__pyx_t_14); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1716; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   __Pyx_GOTREF(__pyx_t_10);
                   __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
                   __pyx_t_17 = Py_TYPE(__pyx_t_10)->tp_iternext;
@@ -54365,7 +54417,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                   __Pyx_GOTREF(__pyx_t_15);
                   index = 1; __pyx_t_2 = __pyx_t_17(__pyx_t_10); if (unlikely(!__pyx_t_2)) goto __pyx_L92_unpacking_failed;
                   __Pyx_GOTREF(__pyx_t_2);
-                  if (__Pyx_IternextUnpackEndCheck(__pyx_t_17(__pyx_t_10), 2) < 0) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1717; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  if (__Pyx_IternextUnpackEndCheck(__pyx_t_17(__pyx_t_10), 2) < 0) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1716; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   __pyx_t_17 = NULL;
                   __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
                   goto __pyx_L93_unpacking_done;
@@ -54373,7 +54425,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                   __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
                   __pyx_t_17 = NULL;
                   if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index);
-                  {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1717; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1716; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   __pyx_L93_unpacking_done:;
                 }
                 __Pyx_XDECREF(__pyx_v_phrase2);
@@ -54383,7 +54435,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 __pyx_v_eindexes = __pyx_t_2;
                 __pyx_t_2 = 0;
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1718
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1717
  *                                     pair_count = 0
  *                                 for phrase2, eindexes in phrase_list:
  *                                     als3 = self.create_alignments(sent_links,num_links,self.findexes,eindexes)             # <<<<<<<<<<<<<<
@@ -54392,31 +54444,31 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
                 __pyx_t_14 = ((PyObject *)__pyx_v_self->findexes);
                 __Pyx_INCREF(__pyx_t_14);
-                __pyx_t_2 = ((PyObject *)((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->create_alignments(__pyx_v_self, __pyx_v_sent_links, __pyx_v_num_links, __pyx_t_14, __pyx_v_eindexes)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1718; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_2 = ((PyObject *)((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->create_alignments(__pyx_v_self, __pyx_v_sent_links, __pyx_v_num_links, __pyx_t_14, __pyx_v_eindexes)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1717; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_2);
                 __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
                 __Pyx_XDECREF(((PyObject *)__pyx_v_als3));
                 __pyx_v_als3 = ((struct __pyx_obj_3_sa_IntList *)__pyx_t_2);
                 __pyx_t_2 = 0;
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1719
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1718
  *                                 for phrase2, eindexes in phrase_list:
  *                                     als3 = self.create_alignments(sent_links,num_links,self.findexes,eindexes)
  *                                     extracts.append((fphr, phrase2, pair_count, tuple(als3)))             # <<<<<<<<<<<<<<
  *                         if (num_gaps < self.max_nonterminals - 1 and
  *                             phrase_len+1 < self.max_length and
  */
-                __pyx_t_2 = PyFloat_FromDouble(__pyx_v_pair_count); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1719; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_2 = PyFloat_FromDouble(__pyx_v_pair_count); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1718; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_2);
-                __pyx_t_14 = PyTuple_New(1); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1719; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_14 = PyTuple_New(1); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1718; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_14);
                 __Pyx_INCREF(((PyObject *)__pyx_v_als3));
                 PyTuple_SET_ITEM(__pyx_t_14, 0, ((PyObject *)__pyx_v_als3));
                 __Pyx_GIVEREF(((PyObject *)__pyx_v_als3));
-                __pyx_t_15 = PyObject_Call(((PyObject *)((PyObject*)(&PyTuple_Type))), ((PyObject *)__pyx_t_14), NULL); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1719; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_15 = PyObject_Call(((PyObject *)((PyObject*)(&PyTuple_Type))), ((PyObject *)__pyx_t_14), NULL); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1718; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_15);
                 __Pyx_DECREF(((PyObject *)__pyx_t_14)); __pyx_t_14 = 0;
-                __pyx_t_14 = PyTuple_New(4); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1719; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_14 = PyTuple_New(4); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1718; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_14);
                 __Pyx_INCREF(((PyObject *)__pyx_v_fphr));
                 PyTuple_SET_ITEM(__pyx_t_14, 0, ((PyObject *)__pyx_v_fphr));
@@ -54430,7 +54482,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 __Pyx_GIVEREF(__pyx_t_15);
                 __pyx_t_2 = 0;
                 __pyx_t_15 = 0;
-                __pyx_t_15 = __Pyx_PyObject_Append(__pyx_v_extracts, ((PyObject *)__pyx_t_14)); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1719; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_15 = __Pyx_PyObject_Append(__pyx_v_extracts, ((PyObject *)__pyx_t_14)); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1718; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_15);
                 __Pyx_DECREF(((PyObject *)__pyx_t_14)); __pyx_t_14 = 0;
                 __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
@@ -54443,7 +54495,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
           }
           __pyx_L79:;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1720
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1719
  *                                     als3 = self.create_alignments(sent_links,num_links,self.findexes,eindexes)
  *                                     extracts.append((fphr, phrase2, pair_count, tuple(als3)))
  *                         if (num_gaps < self.max_nonterminals - 1 and             # <<<<<<<<<<<<<<
@@ -54453,7 +54505,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
           __pyx_t_9 = (__pyx_v_num_gaps < (__pyx_v_self->max_nonterminals - 1));
           if (__pyx_t_9) {
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1721
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1720
  *                                     extracts.append((fphr, phrase2, pair_count, tuple(als3)))
  *                         if (num_gaps < self.max_nonterminals - 1 and
  *                             phrase_len+1 < self.max_length and             # <<<<<<<<<<<<<<
@@ -54463,7 +54515,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             __pyx_t_7 = ((__pyx_v_phrase_len + 1) < __pyx_v_self->max_length);
             if (__pyx_t_7) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1722
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1721
  *                         if (num_gaps < self.max_nonterminals - 1 and
  *                             phrase_len+1 < self.max_length and
  *                             f_back_high == f_high and             # <<<<<<<<<<<<<<
@@ -54473,7 +54525,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
               __pyx_t_18 = (__pyx_v_f_back_high == __pyx_v_f_high);
               if (__pyx_t_18) {
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1723
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1722
  *                             phrase_len+1 < self.max_length and
  *                             f_back_high == f_high and
  *                             f_back_low == f_low and             # <<<<<<<<<<<<<<
@@ -54483,7 +54535,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 __pyx_t_8 = (__pyx_v_f_back_low == __pyx_v_f_low);
                 if (__pyx_t_8) {
 
-                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1724
+                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1723
  *                             f_back_high == f_high and
  *                             f_back_low == f_low and
  *                             f_back_high - f_back_low + (2*self.train_min_gap_size) <= self.train_max_initial_size and             # <<<<<<<<<<<<<<
@@ -54493,7 +54545,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                   __pyx_t_19 = (((__pyx_v_f_back_high - __pyx_v_f_back_low) + (2 * __pyx_v_self->train_min_gap_size)) <= __pyx_v_self->train_max_initial_size);
                   if (__pyx_t_19) {
 
-                    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1725
+                    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1724
  *                             f_back_low == f_low and
  *                             f_back_high - f_back_low + (2*self.train_min_gap_size) <= self.train_max_initial_size and
  *                             f_low >= self.train_min_gap_size and             # <<<<<<<<<<<<<<
@@ -54503,7 +54555,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                     __pyx_t_20 = (__pyx_v_f_low >= __pyx_v_self->train_min_gap_size);
                     if (__pyx_t_20) {
 
-                      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1726
+                      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1725
  *                             f_back_high - f_back_low + (2*self.train_min_gap_size) <= self.train_max_initial_size and
  *                             f_low >= self.train_min_gap_size and
  *                             f_high <= f_sent_len - self.train_min_gap_size and             # <<<<<<<<<<<<<<
@@ -54513,7 +54565,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                       __pyx_t_21 = (__pyx_v_f_high <= (__pyx_v_f_sent_len - __pyx_v_self->train_min_gap_size));
                       if (__pyx_t_21) {
 
-                        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1727
+                        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1726
  *                             f_low >= self.train_min_gap_size and
  *                             f_high <= f_sent_len - self.train_min_gap_size and
  *                             ((not self.tight_phrases) or (f_links_low[f_low-1] != -1 and f_links_low[f_high] != -1))):             # <<<<<<<<<<<<<<
@@ -54563,7 +54615,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
           }
           if (__pyx_t_7) {
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1729
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1728
  *                             ((not self.tight_phrases) or (f_links_low[f_low-1] != -1 and f_links_low[f_high] != -1))):
  * 
  *                             met_constraints = 1             # <<<<<<<<<<<<<<
@@ -54572,7 +54624,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             __pyx_v_met_constraints = 1;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1730
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1729
  * 
  *                             met_constraints = 1
  *                             f_x_low = f_low-self.train_min_gap_size             # <<<<<<<<<<<<<<
@@ -54581,7 +54633,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             __pyx_v_f_x_low = (__pyx_v_f_low - __pyx_v_self->train_min_gap_size);
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1731
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1730
  *                             met_constraints = 1
  *                             f_x_low = f_low-self.train_min_gap_size
  *                             if self.tight_phrases:             # <<<<<<<<<<<<<<
@@ -54590,7 +54642,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             if (__pyx_v_self->tight_phrases) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1732
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1731
  *                             f_x_low = f_low-self.train_min_gap_size
  *                             if self.tight_phrases:
  *                                 while f_x_low >= 0 and f_links_low[f_x_low] == -1:             # <<<<<<<<<<<<<<
@@ -54607,7 +54659,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 }
                 if (!__pyx_t_18) break;
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1733
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1732
  *                             if self.tight_phrases:
  *                                 while f_x_low >= 0 and f_links_low[f_x_low] == -1:
  *                                     f_x_low = f_x_low - 1             # <<<<<<<<<<<<<<
@@ -54620,7 +54672,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             }
             __pyx_L95:;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1734
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1733
  *                                 while f_x_low >= 0 and f_links_low[f_x_low] == -1:
  *                                     f_x_low = f_x_low - 1
  *                             if f_x_low < 0:             # <<<<<<<<<<<<<<
@@ -54630,7 +54682,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             __pyx_t_18 = (__pyx_v_f_x_low < 0);
             if (__pyx_t_18) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1735
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1734
  *                                     f_x_low = f_x_low - 1
  *                             if f_x_low < 0:
  *                                 met_constraints = 0             # <<<<<<<<<<<<<<
@@ -54642,7 +54694,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             }
             __pyx_L98:;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1737
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1736
  *                                 met_constraints = 0
  * 
  *                             f_x_high = f_high+self.train_min_gap_size             # <<<<<<<<<<<<<<
@@ -54651,7 +54703,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             __pyx_v_f_x_high = (__pyx_v_f_high + __pyx_v_self->train_min_gap_size);
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1738
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1737
  * 
  *                             f_x_high = f_high+self.train_min_gap_size
  *                             if self.tight_phrases:             # <<<<<<<<<<<<<<
@@ -54660,7 +54712,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             if (__pyx_v_self->tight_phrases) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1739
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1738
  *                             f_x_high = f_high+self.train_min_gap_size
  *                             if self.tight_phrases:
  *                                 while f_x_high <= f_sent_len and f_links_low[f_x_high-1] == -1:             # <<<<<<<<<<<<<<
@@ -54677,7 +54729,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 }
                 if (!__pyx_t_9) break;
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1740
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1739
  *                             if self.tight_phrases:
  *                                 while f_x_high <= f_sent_len and f_links_low[f_x_high-1] == -1:
  *                                     f_x_high = f_x_high + 1             # <<<<<<<<<<<<<<
@@ -54690,7 +54742,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             }
             __pyx_L99:;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1741
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1740
  *                                 while f_x_high <= f_sent_len and f_links_low[f_x_high-1] == -1:
  *                                     f_x_high = f_x_high + 1
  *                             if f_x_high > f_sent_len or f_x_high - f_x_low > self.train_max_initial_size:             # <<<<<<<<<<<<<<
@@ -54706,7 +54758,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             }
             if (__pyx_t_7) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1742
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1741
  *                                     f_x_high = f_x_high + 1
  *                             if f_x_high > f_sent_len or f_x_high - f_x_low > self.train_max_initial_size:
  *                                 met_constraints = 0             # <<<<<<<<<<<<<<
@@ -54718,7 +54770,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             }
             __pyx_L102:;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1744
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1743
  *                                 met_constraints = 0
  * 
  *                             if (met_constraints and             # <<<<<<<<<<<<<<
@@ -54727,17 +54779,17 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
             if (__pyx_v_met_constraints) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1745
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1744
  * 
  *                             if (met_constraints and
  *                                 self.find_fixpoint(f_x_low, f_x_high,             # <<<<<<<<<<<<<<
  *                                                 f_links_low, f_links_high, e_links_low, e_links_high,
  *                                                 e_low, e_high, &e_x_low, &e_x_high, &f_x_low, &f_x_high,
  */
-              __pyx_t_1 = PyInt_FromLong(__pyx_v_f_x_high); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1745; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_1 = PyInt_FromLong(__pyx_v_f_x_high); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1744; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_1);
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1749
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1748
  *                                                 e_low, e_high, &e_x_low, &e_x_high, &f_x_low, &f_x_high,
  *                                                 f_sent_len, e_sent_len,
  *                                                 self.train_max_initial_size, self.train_max_initial_size,             # <<<<<<<<<<<<<<
@@ -54747,7 +54799,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
               if (((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->find_fixpoint(__pyx_v_self, __pyx_v_f_x_low, __pyx_t_1, __pyx_v_f_links_low, __pyx_v_f_links_high, __pyx_v_e_links_low, __pyx_v_e_links_high, __pyx_v_e_low, __pyx_v_e_high, (&__pyx_v_e_x_low), (&__pyx_v_e_x_high), (&__pyx_v_f_x_low), (&__pyx_v_f_x_high), __pyx_v_f_sent_len, __pyx_v_e_sent_len, __pyx_v_self->train_max_initial_size, __pyx_v_self->train_max_initial_size, 1, 1, 2, 1, 1, 1, 1)) {
                 __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1751
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1750
  *                                                 self.train_max_initial_size, self.train_max_initial_size,
  *                                                 1, 1, 2, 1, 1, 1, 1) and
  *                                 ((not self.tight_phrases) or (f_links_low[f_x_low] != -1 and f_links_low[f_x_high-1] != -1)) and             # <<<<<<<<<<<<<<
@@ -54769,17 +54821,17 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 }
                 if (__pyx_t_9) {
 
-                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1752
+                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1751
  *                                                 1, 1, 2, 1, 1, 1, 1) and
  *                                 ((not self.tight_phrases) or (f_links_low[f_x_low] != -1 and f_links_low[f_x_high-1] != -1)) and
  *                                 self.find_fixpoint(f_x_low, f_low,             # <<<<<<<<<<<<<<
  *                                                 f_links_low, f_links_high, e_links_low, e_links_high,
  *                                                 -1, -1, e_gap_low, e_gap_high, f_gap_low, f_gap_high,
  */
-                  __pyx_t_15 = PyInt_FromLong(__pyx_v_f_low); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1752; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_15 = PyInt_FromLong(__pyx_v_f_low); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1751; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   __Pyx_GOTREF(__pyx_t_15);
 
-                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1756
+                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1755
  *                                                 -1, -1, e_gap_low, e_gap_high, f_gap_low, f_gap_high,
  *                                                 f_sent_len, e_sent_len,
  *                                                 self.train_max_initial_size, self.train_max_initial_size,             # <<<<<<<<<<<<<<
@@ -54790,17 +54842,17 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                   __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
                   if (__pyx_t_3) {
 
-                    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1758
+                    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1757
  *                                                 self.train_max_initial_size, self.train_max_initial_size,
  *                                                 0, 0, 0, 0, 0, 0, 0) and
  *                                 self.find_fixpoint(f_high, f_x_high,             # <<<<<<<<<<<<<<
  *                                                 f_links_low, f_links_high, e_links_low, e_links_high,
  *                                                 -1, -1, e_gap_low+1+num_gaps, e_gap_high+1+num_gaps,
  */
-                    __pyx_t_15 = PyInt_FromLong(__pyx_v_f_x_high); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1758; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                    __pyx_t_15 = PyInt_FromLong(__pyx_v_f_x_high); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1757; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                     __Pyx_GOTREF(__pyx_t_15);
 
-                    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1763
+                    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1762
  *                                                 f_gap_low+1+num_gaps, f_gap_high+1+num_gaps,
  *                                                 f_sent_len, e_sent_len,
  *                                                 self.train_max_initial_size, self.train_max_initial_size,             # <<<<<<<<<<<<<<
@@ -54828,7 +54880,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
             }
             if (__pyx_t_8) {
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1765
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1764
  *                                                 self.train_max_initial_size, self.train_max_initial_size,
  *                                                 0, 0, 0, 0, 0, 0, 0)):
  *                                 fphr_arr._clear()             # <<<<<<<<<<<<<<
@@ -54837,7 +54889,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
               ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_fphr_arr->__pyx_vtab)->_clear(__pyx_v_fphr_arr);
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1766
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1765
  *                                                 0, 0, 0, 0, 0, 0, 0)):
  *                                 fphr_arr._clear()
  *                                 i = 1             # <<<<<<<<<<<<<<
@@ -54846,35 +54898,35 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
               __pyx_v_i = 1;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1767
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1766
  *                                 fphr_arr._clear()
  *                                 i = 1
  *                                 self.findexes.reset()             # <<<<<<<<<<<<<<
  *                                 self.findexes.append(sym_setindex(self.category, i))
  *                                 fphr_arr._append(sym_setindex(self.category, i))
  */
-              __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self->findexes), __pyx_n_s__reset); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1767; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self->findexes), __pyx_n_s__reset); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1766; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_1);
-              __pyx_t_15 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1767; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_15 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1766; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_15);
               __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
               __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1768
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1767
  *                                 i = 1
  *                                 self.findexes.reset()
  *                                 self.findexes.append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
  *                                 fphr_arr._append(sym_setindex(self.category, i))
  *                                 i = i+1
  */
-              __pyx_t_15 = PyInt_FromLong(__pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i)); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1768; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_15 = PyInt_FromLong(__pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i)); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1767; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_15);
-              __pyx_t_1 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes), __pyx_t_15); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1768; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_1 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes), __pyx_t_15); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1767; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_1);
               __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
               __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1769
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1768
  *                                 self.findexes.reset()
  *                                 self.findexes.append(sym_setindex(self.category, i))
  *                                 fphr_arr._append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
@@ -54883,7 +54935,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
               ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_fphr_arr->__pyx_vtab)->_append(__pyx_v_fphr_arr, __pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i));
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1770
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1769
  *                                 self.findexes.append(sym_setindex(self.category, i))
  *                                 fphr_arr._append(sym_setindex(self.category, i))
  *                                 i = i+1             # <<<<<<<<<<<<<<
@@ -54892,27 +54944,27 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
               __pyx_v_i = (__pyx_v_i + 1);
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1771
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1770
  *                                 fphr_arr._append(sym_setindex(self.category, i))
  *                                 i = i+1
  *                                 self.findexes.extend(self.findexes1)             # <<<<<<<<<<<<<<
  *                                 for j from 0 <= j < phrase.n:
  *                                     if sym_isvar(phrase.syms[j]):
  */
-              __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self->findexes), __pyx_n_s__extend); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1771; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self->findexes), __pyx_n_s__extend); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1770; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_1);
-              __pyx_t_15 = PyTuple_New(1); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1771; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_15 = PyTuple_New(1); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1770; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_15);
               __Pyx_INCREF(((PyObject *)__pyx_v_self->findexes1));
               PyTuple_SET_ITEM(__pyx_t_15, 0, ((PyObject *)__pyx_v_self->findexes1));
               __Pyx_GIVEREF(((PyObject *)__pyx_v_self->findexes1));
-              __pyx_t_14 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_15), NULL); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1771; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_14 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_15), NULL); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1770; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_14);
               __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
               __Pyx_DECREF(((PyObject *)__pyx_t_15)); __pyx_t_15 = 0;
               __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1772
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1771
  *                                 i = i+1
  *                                 self.findexes.extend(self.findexes1)
  *                                 for j from 0 <= j < phrase.n:             # <<<<<<<<<<<<<<
@@ -54922,7 +54974,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
               __pyx_t_3 = __pyx_v_phrase->n;
               for (__pyx_v_j = 0; __pyx_v_j < __pyx_t_3; __pyx_v_j++) {
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1773
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1772
  *                                 self.findexes.extend(self.findexes1)
  *                                 for j from 0 <= j < phrase.n:
  *                                     if sym_isvar(phrase.syms[j]):             # <<<<<<<<<<<<<<
@@ -54932,7 +54984,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 __pyx_t_4 = __pyx_f_3_sa_sym_isvar((__pyx_v_phrase->syms[__pyx_v_j]));
                 if (__pyx_t_4) {
 
-                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1774
+                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1773
  *                                 for j from 0 <= j < phrase.n:
  *                                     if sym_isvar(phrase.syms[j]):
  *                                         fphr_arr._append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
@@ -54941,7 +54993,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
                   ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_fphr_arr->__pyx_vtab)->_append(__pyx_v_fphr_arr, __pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i));
 
-                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1775
+                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1774
  *                                     if sym_isvar(phrase.syms[j]):
  *                                         fphr_arr._append(sym_setindex(self.category, i))
  *                                         i = i + 1             # <<<<<<<<<<<<<<
@@ -54953,7 +55005,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 }
                 /*else*/ {
 
-                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1777
+                  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1776
  *                                         i = i + 1
  *                                     else:
  *                                         fphr_arr._append(phrase.syms[j])             # <<<<<<<<<<<<<<
@@ -54965,7 +55017,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 __pyx_L106:;
               }
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1778
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1777
  *                                     else:
  *                                         fphr_arr._append(phrase.syms[j])
  *                                 fphr_arr._append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
@@ -54974,81 +55026,81 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
               ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_fphr_arr->__pyx_vtab)->_append(__pyx_v_fphr_arr, __pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i));
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1779
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1778
  *                                         fphr_arr._append(phrase.syms[j])
  *                                 fphr_arr._append(sym_setindex(self.category, i))
  *                                 self.findexes.append(sym_setindex(self.category, i))             # <<<<<<<<<<<<<<
  *                                 fphr = Phrase(fphr_arr)
  *                                 phrase_list = self.extract_phrases(e_x_low, e_x_high, e_gap_low, e_gap_high, e_links_low, num_gaps+2,
  */
-              __pyx_t_14 = PyInt_FromLong(__pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i)); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1779; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_14 = PyInt_FromLong(__pyx_f_3_sa_sym_setindex(__pyx_v_self->category, __pyx_v_i)); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1778; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_14);
-              __pyx_t_15 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes), __pyx_t_14); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1779; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_15 = __Pyx_PyObject_Append(((PyObject *)__pyx_v_self->findexes), __pyx_t_14); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1778; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_15);
               __Pyx_DECREF(__pyx_t_14); __pyx_t_14 = 0;
               __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1780
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1779
  *                                 fphr_arr._append(sym_setindex(self.category, i))
  *                                 self.findexes.append(sym_setindex(self.category, i))
  *                                 fphr = Phrase(fphr_arr)             # <<<<<<<<<<<<<<
  *                                 phrase_list = self.extract_phrases(e_x_low, e_x_high, e_gap_low, e_gap_high, e_links_low, num_gaps+2,
  *                                                     f_x_low, f_x_high, f_gap_low, f_gap_high, f_links_low,
  */
-              __pyx_t_15 = PyTuple_New(1); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1780; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_15 = PyTuple_New(1); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1779; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_15);
               __Pyx_INCREF(((PyObject *)__pyx_v_fphr_arr));
               PyTuple_SET_ITEM(__pyx_t_15, 0, ((PyObject *)__pyx_v_fphr_arr));
               __Pyx_GIVEREF(((PyObject *)__pyx_v_fphr_arr));
-              __pyx_t_14 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_Phrase)), ((PyObject *)__pyx_t_15), NULL); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1780; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_14 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_Phrase)), ((PyObject *)__pyx_t_15), NULL); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1779; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_14);
               __Pyx_DECREF(((PyObject *)__pyx_t_15)); __pyx_t_15 = 0;
               __Pyx_DECREF(((PyObject *)__pyx_v_fphr));
               __pyx_v_fphr = ((struct __pyx_obj_3_sa_Phrase *)__pyx_t_14);
               __pyx_t_14 = 0;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1783
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1782
  *                                 phrase_list = self.extract_phrases(e_x_low, e_x_high, e_gap_low, e_gap_high, e_links_low, num_gaps+2,
  *                                                     f_x_low, f_x_high, f_gap_low, f_gap_high, f_links_low,
  *                                                     matching.sent_id, e_sent_len, e_sent_start)             # <<<<<<<<<<<<<<
  *                                 if len(phrase_list) > 0:
  *                                     pair_count = 1.0 / len(phrase_list)
  */
-              __pyx_t_14 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->extract_phrases(__pyx_v_self, __pyx_v_e_x_low, __pyx_v_e_x_high, __pyx_v_e_gap_low, __pyx_v_e_gap_high, __pyx_v_e_links_low, (__pyx_v_num_gaps + 2), __pyx_v_f_x_low, __pyx_v_f_x_high, __pyx_v_f_gap_low, __pyx_v_f_gap_high, __pyx_v_f_links_low, __pyx_v_matching->sent_id, __pyx_v_e_sent_len, __pyx_v_e_sent_start); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1781; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_14 = ((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->extract_phrases(__pyx_v_self, __pyx_v_e_x_low, __pyx_v_e_x_high, __pyx_v_e_gap_low, __pyx_v_e_gap_high, __pyx_v_e_links_low, (__pyx_v_num_gaps + 2), __pyx_v_f_x_low, __pyx_v_f_x_high, __pyx_v_f_gap_low, __pyx_v_f_gap_high, __pyx_v_f_links_low, __pyx_v_matching->sent_id, __pyx_v_e_sent_len, __pyx_v_e_sent_start); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1780; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __Pyx_GOTREF(__pyx_t_14);
               __Pyx_XDECREF(__pyx_v_phrase_list);
               __pyx_v_phrase_list = __pyx_t_14;
               __pyx_t_14 = 0;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1784
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1783
  *                                                     f_x_low, f_x_high, f_gap_low, f_gap_high, f_links_low,
  *                                                     matching.sent_id, e_sent_len, e_sent_start)
  *                                 if len(phrase_list) > 0:             # <<<<<<<<<<<<<<
  *                                     pair_count = 1.0 / len(phrase_list)
  *                                 else:
  */
-              __pyx_t_13 = PyObject_Length(__pyx_v_phrase_list); if (unlikely(__pyx_t_13 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1784; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+              __pyx_t_13 = PyObject_Length(__pyx_v_phrase_list); if (unlikely(__pyx_t_13 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1783; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
               __pyx_t_8 = (__pyx_t_13 > 0);
               if (__pyx_t_8) {
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1785
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1784
  *                                                     matching.sent_id, e_sent_len, e_sent_start)
  *                                 if len(phrase_list) > 0:
  *                                     pair_count = 1.0 / len(phrase_list)             # <<<<<<<<<<<<<<
  *                                 else:
  *                                     pair_count = 0
  */
-                __pyx_t_13 = PyObject_Length(__pyx_v_phrase_list); if (unlikely(__pyx_t_13 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1785; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_13 = PyObject_Length(__pyx_v_phrase_list); if (unlikely(__pyx_t_13 == -1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1784; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 if (unlikely(__pyx_t_13 == 0)) {
                   PyErr_Format(PyExc_ZeroDivisionError, "float division");
-                  {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1785; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1784; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 }
                 __pyx_v_pair_count = (1.0 / __pyx_t_13);
                 goto __pyx_L107;
               }
               /*else*/ {
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1787
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1786
  *                                     pair_count = 1.0 / len(phrase_list)
  *                                 else:
  *                                     pair_count = 0             # <<<<<<<<<<<<<<
@@ -55059,7 +55111,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
               }
               __pyx_L107:;
 
-              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1788
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1787
  *                                 else:
  *                                     pair_count = 0
  *                                 for phrase2, eindexes in phrase_list:             # <<<<<<<<<<<<<<
@@ -55070,7 +55122,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 __pyx_t_14 = __pyx_v_phrase_list; __Pyx_INCREF(__pyx_t_14); __pyx_t_13 = 0;
                 __pyx_t_16 = NULL;
               } else {
-                __pyx_t_13 = -1; __pyx_t_14 = PyObject_GetIter(__pyx_v_phrase_list); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1788; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_13 = -1; __pyx_t_14 = PyObject_GetIter(__pyx_v_phrase_list); if (unlikely(!__pyx_t_14)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1787; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_14);
                 __pyx_t_16 = Py_TYPE(__pyx_t_14)->tp_iternext;
               }
@@ -55078,23 +55130,23 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 if (!__pyx_t_16 && PyList_CheckExact(__pyx_t_14)) {
                   if (__pyx_t_13 >= PyList_GET_SIZE(__pyx_t_14)) break;
                   #if CYTHON_COMPILING_IN_CPYTHON
-                  __pyx_t_15 = PyList_GET_ITEM(__pyx_t_14, __pyx_t_13); __Pyx_INCREF(__pyx_t_15); __pyx_t_13++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1788; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_15 = PyList_GET_ITEM(__pyx_t_14, __pyx_t_13); __Pyx_INCREF(__pyx_t_15); __pyx_t_13++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1787; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   #else
-                  __pyx_t_15 = PySequence_ITEM(__pyx_t_14, __pyx_t_13); __pyx_t_13++; if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1788; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_15 = PySequence_ITEM(__pyx_t_14, __pyx_t_13); __pyx_t_13++; if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1787; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   #endif
                 } else if (!__pyx_t_16 && PyTuple_CheckExact(__pyx_t_14)) {
                   if (__pyx_t_13 >= PyTuple_GET_SIZE(__pyx_t_14)) break;
                   #if CYTHON_COMPILING_IN_CPYTHON
-                  __pyx_t_15 = PyTuple_GET_ITEM(__pyx_t_14, __pyx_t_13); __Pyx_INCREF(__pyx_t_15); __pyx_t_13++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1788; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_15 = PyTuple_GET_ITEM(__pyx_t_14, __pyx_t_13); __Pyx_INCREF(__pyx_t_15); __pyx_t_13++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1787; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   #else
-                  __pyx_t_15 = PySequence_ITEM(__pyx_t_14, __pyx_t_13); __pyx_t_13++; if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1788; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_15 = PySequence_ITEM(__pyx_t_14, __pyx_t_13); __pyx_t_13++; if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1787; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   #endif
                 } else {
                   __pyx_t_15 = __pyx_t_16(__pyx_t_14);
                   if (unlikely(!__pyx_t_15)) {
                     if (PyErr_Occurred()) {
                       if (likely(PyErr_ExceptionMatches(PyExc_StopIteration))) PyErr_Clear();
-                      else {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1788; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                      else {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1787; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                     }
                     break;
                   }
@@ -55110,7 +55162,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                   if (unlikely(size != 2)) {
                     if (size > 2) __Pyx_RaiseTooManyValuesError(2);
                     else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size);
-                    {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1788; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                    {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1787; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   }
                   #if CYTHON_COMPILING_IN_CPYTHON
                   if (likely(PyTuple_CheckExact(sequence))) {
@@ -55123,14 +55175,14 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                   __Pyx_INCREF(__pyx_t_1);
                   __Pyx_INCREF(__pyx_t_2);
                   #else
-                  __pyx_t_1 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1788; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-                  __pyx_t_2 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1788; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_1 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1787; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_2 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1787; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   #endif
                   __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
                 } else
                 {
                   Py_ssize_t index = -1;
-                  __pyx_t_10 = PyObject_GetIter(__pyx_t_15); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1788; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  __pyx_t_10 = PyObject_GetIter(__pyx_t_15); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1787; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   __Pyx_GOTREF(__pyx_t_10);
                   __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
                   __pyx_t_17 = Py_TYPE(__pyx_t_10)->tp_iternext;
@@ -55138,7 +55190,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                   __Pyx_GOTREF(__pyx_t_1);
                   index = 1; __pyx_t_2 = __pyx_t_17(__pyx_t_10); if (unlikely(!__pyx_t_2)) goto __pyx_L110_unpacking_failed;
                   __Pyx_GOTREF(__pyx_t_2);
-                  if (__Pyx_IternextUnpackEndCheck(__pyx_t_17(__pyx_t_10), 2) < 0) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1788; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  if (__Pyx_IternextUnpackEndCheck(__pyx_t_17(__pyx_t_10), 2) < 0) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1787; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   __pyx_t_17 = NULL;
                   __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
                   goto __pyx_L111_unpacking_done;
@@ -55146,7 +55198,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                   __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
                   __pyx_t_17 = NULL;
                   if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index);
-                  {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1788; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                  {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1787; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                   __pyx_L111_unpacking_done:;
                 }
                 __Pyx_XDECREF(__pyx_v_phrase2);
@@ -55156,7 +55208,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 __pyx_v_eindexes = __pyx_t_2;
                 __pyx_t_2 = 0;
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1789
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1788
  *                                     pair_count = 0
  *                                 for phrase2, eindexes in phrase_list:
  *                                     als4 = self.create_alignments(sent_links,num_links,self.findexes,eindexes)             # <<<<<<<<<<<<<<
@@ -55165,31 +55217,31 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
                 __pyx_t_15 = ((PyObject *)__pyx_v_self->findexes);
                 __Pyx_INCREF(__pyx_t_15);
-                __pyx_t_2 = ((PyObject *)((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->create_alignments(__pyx_v_self, __pyx_v_sent_links, __pyx_v_num_links, __pyx_t_15, __pyx_v_eindexes)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1789; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_2 = ((PyObject *)((struct __pyx_vtabstruct_3_sa_HieroCachingRuleFactory *)__pyx_v_self->__pyx_vtab)->create_alignments(__pyx_v_self, __pyx_v_sent_links, __pyx_v_num_links, __pyx_t_15, __pyx_v_eindexes)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1788; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_2);
                 __Pyx_DECREF(__pyx_t_15); __pyx_t_15 = 0;
                 __Pyx_XDECREF(((PyObject *)__pyx_v_als4));
                 __pyx_v_als4 = ((struct __pyx_obj_3_sa_IntList *)__pyx_t_2);
                 __pyx_t_2 = 0;
 
-                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1790
+                /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1789
  *                                 for phrase2, eindexes in phrase_list:
  *                                     als4 = self.create_alignments(sent_links,num_links,self.findexes,eindexes)
  *                                     extracts.append((fphr, phrase2, pair_count, tuple(als4)))             # <<<<<<<<<<<<<<
  *             else:
  *                 reason_for_failure = "Unable to extract basic phrase"
  */
-                __pyx_t_2 = PyFloat_FromDouble(__pyx_v_pair_count); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1790; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_2 = PyFloat_FromDouble(__pyx_v_pair_count); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1789; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_2);
-                __pyx_t_15 = PyTuple_New(1); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1790; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_15 = PyTuple_New(1); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1789; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_15);
                 __Pyx_INCREF(((PyObject *)__pyx_v_als4));
                 PyTuple_SET_ITEM(__pyx_t_15, 0, ((PyObject *)__pyx_v_als4));
                 __Pyx_GIVEREF(((PyObject *)__pyx_v_als4));
-                __pyx_t_1 = PyObject_Call(((PyObject *)((PyObject*)(&PyTuple_Type))), ((PyObject *)__pyx_t_15), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1790; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_1 = PyObject_Call(((PyObject *)((PyObject*)(&PyTuple_Type))), ((PyObject *)__pyx_t_15), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1789; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_1);
                 __Pyx_DECREF(((PyObject *)__pyx_t_15)); __pyx_t_15 = 0;
-                __pyx_t_15 = PyTuple_New(4); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1790; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_15 = PyTuple_New(4); if (unlikely(!__pyx_t_15)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1789; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_15);
                 __Pyx_INCREF(((PyObject *)__pyx_v_fphr));
                 PyTuple_SET_ITEM(__pyx_t_15, 0, ((PyObject *)__pyx_v_fphr));
@@ -55203,7 +55255,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
                 __Pyx_GIVEREF(__pyx_t_1);
                 __pyx_t_2 = 0;
                 __pyx_t_1 = 0;
-                __pyx_t_1 = __Pyx_PyObject_Append(__pyx_v_extracts, ((PyObject *)__pyx_t_15)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1790; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+                __pyx_t_1 = __Pyx_PyObject_Append(__pyx_v_extracts, ((PyObject *)__pyx_t_15)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1789; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
                 __Pyx_GOTREF(__pyx_t_1);
                 __Pyx_DECREF(((PyObject *)__pyx_t_15)); __pyx_t_15 = 0;
                 __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
@@ -55225,7 +55277,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
     }
     /*else*/ {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1792
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1791
  *                                     extracts.append((fphr, phrase2, pair_count, tuple(als4)))
  *             else:
  *                 reason_for_failure = "Unable to extract basic phrase"             # <<<<<<<<<<<<<<
@@ -55241,7 +55293,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
   }
   __pyx_L33:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1794
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1793
  *                 reason_for_failure = "Unable to extract basic phrase"
  * 
  *         free(sent_links)             # <<<<<<<<<<<<<<
@@ -55250,7 +55302,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   free(__pyx_v_sent_links);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1795
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1794
  * 
  *         free(sent_links)
  *         free(f_links_low)             # <<<<<<<<<<<<<<
@@ -55259,7 +55311,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   free(__pyx_v_f_links_low);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1796
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1795
  *         free(sent_links)
  *         free(f_links_low)
  *         free(f_links_high)             # <<<<<<<<<<<<<<
@@ -55268,7 +55320,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   free(__pyx_v_f_links_high);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1797
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1796
  *         free(f_links_low)
  *         free(f_links_high)
  *         free(e_links_low)             # <<<<<<<<<<<<<<
@@ -55277,7 +55329,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   free(__pyx_v_e_links_low);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1798
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1797
  *         free(f_links_high)
  *         free(e_links_low)
  *         free(e_links_high)             # <<<<<<<<<<<<<<
@@ -55286,7 +55338,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   free(__pyx_v_e_links_high);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1799
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1798
  *         free(e_links_low)
  *         free(e_links_high)
  *         free(f_gap_low)             # <<<<<<<<<<<<<<
@@ -55295,7 +55347,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   free(__pyx_v_f_gap_low);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1800
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1799
  *         free(e_links_high)
  *         free(f_gap_low)
  *         free(f_gap_high)             # <<<<<<<<<<<<<<
@@ -55304,7 +55356,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   free(__pyx_v_f_gap_high);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1801
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1800
  *         free(f_gap_low)
  *         free(f_gap_high)
  *         free(e_gap_low)             # <<<<<<<<<<<<<<
@@ -55313,7 +55365,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   free(__pyx_v_e_gap_low);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1802
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1801
  *         free(f_gap_high)
  *         free(e_gap_low)
  *         free(e_gap_high)             # <<<<<<<<<<<<<<
@@ -55322,7 +55374,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  */
   free(__pyx_v_e_gap_high);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1804
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1803
  *         free(e_gap_high)
  * 
  *         return extracts             # <<<<<<<<<<<<<<
@@ -57566,24 +57618,41 @@ static int __pyx_tp_clear_3_sa_DataArray(PyObject *o) {
   return 0;
 }
 
+static PyObject *__pyx_getprop_3_sa_9DataArray_data(PyObject *o, CYTHON_UNUSED void *x) {
+  return __pyx_pw_3_sa_9DataArray_4data_1__get__(o);
+}
+
+static int __pyx_setprop_3_sa_9DataArray_data(PyObject *o, PyObject *v, CYTHON_UNUSED void *x) {
+  if (v) {
+    return __pyx_pw_3_sa_9DataArray_4data_3__set__(o, v);
+  }
+  else {
+    return __pyx_pw_3_sa_9DataArray_4data_5__del__(o);
+  }
+}
+
 static PyMethodDef __pyx_methods_3_sa_DataArray[] = {
-  {__Pyx_NAMESTR("get_data"), (PyCFunction)__pyx_pw_3_sa_9DataArray_5get_data, METH_NOARGS, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("get_sentence_id"), (PyCFunction)__pyx_pw_3_sa_9DataArray_7get_sentence_id, METH_O, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("get_sentence"), (PyCFunction)__pyx_pw_3_sa_9DataArray_9get_sentence, METH_O, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("get_sentence_position"), (PyCFunction)__pyx_pw_3_sa_9DataArray_11get_sentence_position, METH_O, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("get_id"), (PyCFunction)__pyx_pw_3_sa_9DataArray_13get_id, METH_O, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("get_word"), (PyCFunction)__pyx_pw_3_sa_9DataArray_15get_word, METH_O, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("write_text"), (PyCFunction)__pyx_pw_3_sa_9DataArray_17write_text, METH_O, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("read_text"), (PyCFunction)__pyx_pw_3_sa_9DataArray_19read_text, METH_O, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("read_bitext"), (PyCFunction)__pyx_pw_3_sa_9DataArray_21read_bitext, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("read_text_data"), (PyCFunction)__pyx_pw_3_sa_9DataArray_23read_text_data, METH_O, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("read_binary"), (PyCFunction)__pyx_pw_3_sa_9DataArray_25read_binary, METH_O, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("write_binary"), (PyCFunction)__pyx_pw_3_sa_9DataArray_27write_binary, METH_O, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("write_enhanced_handle"), (PyCFunction)__pyx_pw_3_sa_9DataArray_29write_enhanced_handle, METH_O, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("write_enhanced"), (PyCFunction)__pyx_pw_3_sa_9DataArray_31write_enhanced, METH_O, __Pyx_DOCSTR(0)},
+  {__Pyx_NAMESTR("get_sentence_id"), (PyCFunction)__pyx_pw_3_sa_9DataArray_5get_sentence_id, METH_O, __Pyx_DOCSTR(0)},
+  {__Pyx_NAMESTR("get_sentence"), (PyCFunction)__pyx_pw_3_sa_9DataArray_7get_sentence, METH_O, __Pyx_DOCSTR(0)},
+  {__Pyx_NAMESTR("get_sentence_position"), (PyCFunction)__pyx_pw_3_sa_9DataArray_9get_sentence_position, METH_O, __Pyx_DOCSTR(0)},
+  {__Pyx_NAMESTR("get_id"), (PyCFunction)__pyx_pw_3_sa_9DataArray_11get_id, METH_O, __Pyx_DOCSTR(0)},
+  {__Pyx_NAMESTR("get_word"), (PyCFunction)__pyx_pw_3_sa_9DataArray_13get_word, METH_O, __Pyx_DOCSTR(0)},
+  {__Pyx_NAMESTR("write_text"), (PyCFunction)__pyx_pw_3_sa_9DataArray_15write_text, METH_O, __Pyx_DOCSTR(0)},
+  {__Pyx_NAMESTR("read_text"), (PyCFunction)__pyx_pw_3_sa_9DataArray_17read_text, METH_O, __Pyx_DOCSTR(0)},
+  {__Pyx_NAMESTR("read_bitext"), (PyCFunction)__pyx_pw_3_sa_9DataArray_19read_bitext, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(0)},
+  {__Pyx_NAMESTR("read_text_data"), (PyCFunction)__pyx_pw_3_sa_9DataArray_21read_text_data, METH_O, __Pyx_DOCSTR(0)},
+  {__Pyx_NAMESTR("read_binary"), (PyCFunction)__pyx_pw_3_sa_9DataArray_23read_binary, METH_O, __Pyx_DOCSTR(0)},
+  {__Pyx_NAMESTR("write_binary"), (PyCFunction)__pyx_pw_3_sa_9DataArray_25write_binary, METH_O, __Pyx_DOCSTR(0)},
+  {__Pyx_NAMESTR("write_enhanced_handle"), (PyCFunction)__pyx_pw_3_sa_9DataArray_27write_enhanced_handle, METH_O, __Pyx_DOCSTR(0)},
+  {__Pyx_NAMESTR("write_enhanced"), (PyCFunction)__pyx_pw_3_sa_9DataArray_29write_enhanced, METH_O, __Pyx_DOCSTR(0)},
   {0, 0, 0, 0}
 };
 
+static struct PyGetSetDef __pyx_getsets_3_sa_DataArray[] = {
+  {(char *)"data", __pyx_getprop_3_sa_9DataArray_data, __pyx_setprop_3_sa_9DataArray_data, 0, 0},
+  {0, 0, 0, 0, 0}
+};
+
 static PyNumberMethods __pyx_tp_as_number_DataArray = {
   0, /*nb_add*/
   0, /*nb_subtract*/
@@ -57716,7 +57785,7 @@ static PyTypeObject __pyx_type_3_sa_DataArray = {
   0, /*tp_iternext*/
   __pyx_methods_3_sa_DataArray, /*tp_methods*/
   0, /*tp_members*/
-  0, /*tp_getset*/
+  __pyx_getsets_3_sa_DataArray, /*tp_getset*/
   0, /*tp_base*/
   0, /*tp_dict*/
   0, /*tp_descr_get*/
@@ -66007,6 +66076,7 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = {
   {&__pyx_n_s__fphrase, __pyx_k__fphrase, sizeof(__pyx_k__fphrase), 0, 0, 1, 1},
   {&__pyx_n_s__from_binary, __pyx_k__from_binary, sizeof(__pyx_k__from_binary), 0, 0, 1, 1},
   {&__pyx_n_s__from_data, __pyx_k__from_data, sizeof(__pyx_k__from_data), 0, 0, 1, 1},
+  {&__pyx_n_s__from_iterable, __pyx_k__from_iterable, sizeof(__pyx_k__from_iterable), 0, 0, 1, 1},
   {&__pyx_n_s__from_stats, __pyx_k__from_stats, sizeof(__pyx_k__from_stats), 0, 0, 1, 1},
   {&__pyx_n_s__from_text, __pyx_k__from_text, sizeof(__pyx_k__from_text), 0, 0, 1, 1},
   {&__pyx_n_s__frontier, __pyx_k__frontier, sizeof(__pyx_k__frontier), 0, 0, 1, 1},
@@ -66145,7 +66215,7 @@ static int __Pyx_InitCachedBuiltins(void) {
   __pyx_builtin_IndexError = __Pyx_GetName(__pyx_b, __pyx_n_s__IndexError); if (!__pyx_builtin_IndexError) {__pyx_filename = __pyx_f[1]; __pyx_lineno = 28; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_builtin_range = __Pyx_GetName(__pyx_b, __pyx_n_s__range); if (!__pyx_builtin_range) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 23; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_builtin_TypeError = __Pyx_GetName(__pyx_b, __pyx_n_s__TypeError); if (!__pyx_builtin_TypeError) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 109; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __pyx_builtin_enumerate = __Pyx_GetName(__pyx_b, __pyx_n_s__enumerate); if (!__pyx_builtin_enumerate) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 81; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_builtin_enumerate = __Pyx_GetName(__pyx_b, __pyx_n_s__enumerate); if (!__pyx_builtin_enumerate) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 78; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_builtin_map = __Pyx_GetName(__pyx_b, __pyx_n_s__map); if (!__pyx_builtin_map) {__pyx_filename = __pyx_f[4]; __pyx_lineno = 59; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_builtin_Exception = __Pyx_GetName(__pyx_b, __pyx_n_s__Exception); if (!__pyx_builtin_Exception) {__pyx_filename = __pyx_f[5]; __pyx_lineno = 123; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_builtin_zip = __Pyx_GetName(__pyx_b, __pyx_n_s__zip); if (!__pyx_builtin_zip) {__pyx_filename = __pyx_f[5]; __pyx_lineno = 363; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
@@ -66213,28 +66283,28 @@ static int __Pyx_InitCachedConstants(void) {
   __Pyx_GIVEREF(__pyx_int_1000);
   __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_12));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":68
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":65
  *                     f.write("%s " % self.get_word(w_id))
  *                 if w_id == 1:
  *                     f.write("\n")             # <<<<<<<<<<<<<<
  * 
  *     def read_text(self, char* filename):
  */
-  __pyx_k_tuple_15 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_15)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_k_tuple_15 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_15)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 65; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_k_tuple_15);
   __Pyx_INCREF(((PyObject *)__pyx_kp_s_14));
   PyTuple_SET_ITEM(__pyx_k_tuple_15, 0, ((PyObject *)__pyx_kp_s_14));
   __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_14));
   __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_15));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":63
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":60
  * 
  *     def write_text(self, char* filename):
  *         with open(filename, "w") as f:             # <<<<<<<<<<<<<<
  *             for w_id in self.data:
  *                 if w_id > 1:
  */
-  __pyx_k_tuple_16 = PyTuple_New(3); if (unlikely(!__pyx_k_tuple_16)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 63; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_k_tuple_16 = PyTuple_New(3); if (unlikely(!__pyx_k_tuple_16)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 60; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_k_tuple_16);
   __Pyx_INCREF(Py_None);
   PyTuple_SET_ITEM(__pyx_k_tuple_16, 0, Py_None);
@@ -66247,14 +66317,14 @@ static int __Pyx_InitCachedConstants(void) {
   __Pyx_GIVEREF(Py_None);
   __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_16));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":71
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":68
  * 
  *     def read_text(self, char* filename):
  *         with gzip_or_text(filename) as fp:             # <<<<<<<<<<<<<<
  *             self.read_text_data(fp)
  * 
  */
-  __pyx_k_tuple_17 = PyTuple_New(3); if (unlikely(!__pyx_k_tuple_17)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_k_tuple_17 = PyTuple_New(3); if (unlikely(!__pyx_k_tuple_17)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_k_tuple_17);
   __Pyx_INCREF(Py_None);
   PyTuple_SET_ITEM(__pyx_k_tuple_17, 0, Py_None);
@@ -66267,28 +66337,28 @@ static int __Pyx_InitCachedConstants(void) {
   __Pyx_GIVEREF(Py_None);
   __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_17));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":76
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":73
  *     def read_bitext(self, char* filename, int side):
  *         with gzip_or_text(filename) as fp:
  *             data = (line.split(' ||| ')[side] for line in fp)             # <<<<<<<<<<<<<<
  *             self.read_text_data(data)
  * 
  */
-  __pyx_k_tuple_19 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_19)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 76; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_k_tuple_19 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_19)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_k_tuple_19);
   __Pyx_INCREF(((PyObject *)__pyx_kp_s_18));
   PyTuple_SET_ITEM(__pyx_k_tuple_19, 0, ((PyObject *)__pyx_kp_s_18));
   __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_18));
   __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_19));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":75
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":72
  * 
  *     def read_bitext(self, char* filename, int side):
  *         with gzip_or_text(filename) as fp:             # <<<<<<<<<<<<<<
  *             data = (line.split(' ||| ')[side] for line in fp)
  *             self.read_text_data(data)
  */
-  __pyx_k_tuple_20 = PyTuple_New(3); if (unlikely(!__pyx_k_tuple_20)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 75; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_k_tuple_20 = PyTuple_New(3); if (unlikely(!__pyx_k_tuple_20)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_k_tuple_20);
   __Pyx_INCREF(Py_None);
   PyTuple_SET_ITEM(__pyx_k_tuple_20, 0, Py_None);
@@ -66301,69 +66371,69 @@ static int __Pyx_InitCachedConstants(void) {
   __Pyx_GIVEREF(Py_None);
   __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_20));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":146
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":143
  *         for i in self.data:
  *             f.write("%d " %i)
  *         f.write("\n")             # <<<<<<<<<<<<<<
  *         for i in self.sent_index:
  *             f.write("%d " %i)
  */
-  __pyx_k_tuple_22 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_22)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 146; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_k_tuple_22 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_22)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 143; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_k_tuple_22);
   __Pyx_INCREF(((PyObject *)__pyx_kp_s_14));
   PyTuple_SET_ITEM(__pyx_k_tuple_22, 0, ((PyObject *)__pyx_kp_s_14));
   __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_14));
   __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_22));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":149
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":146
  *         for i in self.sent_index:
  *             f.write("%d " %i)
  *         f.write("\n")             # <<<<<<<<<<<<<<
  *         for i in self.sent_id:
  *             f.write("%d " %i)
  */
-  __pyx_k_tuple_23 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_23)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 149; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_k_tuple_23 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_23)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 146; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_k_tuple_23);
   __Pyx_INCREF(((PyObject *)__pyx_kp_s_14));
   PyTuple_SET_ITEM(__pyx_k_tuple_23, 0, ((PyObject *)__pyx_kp_s_14));
   __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_14));
   __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_23));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":152
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":149
  *         for i in self.sent_id:
  *             f.write("%d " %i)
  *         f.write("\n")             # <<<<<<<<<<<<<<
  *         for word in self.id2word:
  *             f.write("%s %d " % (word, self.word2id[word]))
  */
-  __pyx_k_tuple_24 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_24)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_k_tuple_24 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_24)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 149; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_k_tuple_24);
   __Pyx_INCREF(((PyObject *)__pyx_kp_s_14));
   PyTuple_SET_ITEM(__pyx_k_tuple_24, 0, ((PyObject *)__pyx_kp_s_14));
   __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_14));
   __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_24));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":155
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":152
  *         for word in self.id2word:
  *             f.write("%s %d " % (word, self.word2id[word]))
  *         f.write("\n")             # <<<<<<<<<<<<<<
  * 
  *     def write_enhanced(self, char* filename):
  */
-  __pyx_k_tuple_26 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_26)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 155; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_k_tuple_26 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_26)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_k_tuple_26);
   __Pyx_INCREF(((PyObject *)__pyx_kp_s_14));
   PyTuple_SET_ITEM(__pyx_k_tuple_26, 0, ((PyObject *)__pyx_kp_s_14));
   __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_14));
   __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_26));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":158
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":155
  * 
  *     def write_enhanced(self, char* filename):
  *         with open(filename, "w") as f:             # <<<<<<<<<<<<<<
  *             self.write_enhanced_handle(self, f)
  */
-  __pyx_k_tuple_28 = PyTuple_New(3); if (unlikely(!__pyx_k_tuple_28)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 158; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_k_tuple_28 = PyTuple_New(3); if (unlikely(!__pyx_k_tuple_28)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 155; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_k_tuple_28);
   __Pyx_INCREF(Py_None);
   PyTuple_SET_ITEM(__pyx_k_tuple_28, 0, Py_None);
@@ -67280,9 +67350,9 @@ PyMODINIT_FUNC PyInit__sa(void)
   __pyx_ptype_3_sa_Scorer = &__pyx_type_3_sa_Scorer;
   if (PyType_Ready(&__pyx_type_3_sa___pyx_scope_struct____iter__) < 0) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 81; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_ptype_3_sa___pyx_scope_struct____iter__ = &__pyx_type_3_sa___pyx_scope_struct____iter__;
-  if (PyType_Ready(&__pyx_type_3_sa___pyx_scope_struct_1_read_bitext) < 0) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 74; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyType_Ready(&__pyx_type_3_sa___pyx_scope_struct_1_read_bitext) < 0) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 71; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_ptype_3_sa___pyx_scope_struct_1_read_bitext = &__pyx_type_3_sa___pyx_scope_struct_1_read_bitext;
-  if (PyType_Ready(&__pyx_type_3_sa___pyx_scope_struct_2_genexpr) < 0) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 76; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyType_Ready(&__pyx_type_3_sa___pyx_scope_struct_2_genexpr) < 0) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_ptype_3_sa___pyx_scope_struct_2_genexpr = &__pyx_type_3_sa___pyx_scope_struct_2_genexpr;
   if (PyType_Ready(&__pyx_type_3_sa___pyx_scope_struct_3_compute_stats) < 0) {__pyx_filename = __pyx_f[9]; __pyx_lineno = 36; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_ptype_3_sa___pyx_scope_struct_3_compute_stats = &__pyx_type_3_sa___pyx_scope_struct_3_compute_stats;
diff --git a/python/src/sa/data_array.pxi b/python/src/sa/data_array.pxi
index ce2d3a79..371708c5 100644
--- a/python/src/sa/data_array.pxi
+++ b/python/src/sa/data_array.pxi
@@ -9,8 +9,8 @@ from libc.string cimport memset, strcpy
 cdef class DataArray:
     cdef word2id
     cdef id2word
-    cdef IntList data
-    cdef IntList sent_id
+    cdef public IntList data
+    cdef public IntList sent_id
     cdef IntList sent_index
     cdef bint use_sent_id
 
@@ -32,9 +32,6 @@ cdef class DataArray:
     def __len__(self):
         return len(self.data)
 
-    def get_data(self):
-        return self.data
-
     def get_sentence_id(self, i):
         return self.sent_id.arr[i]
 
diff --git a/python/src/sa/rulefactory.pxi b/python/src/sa/rulefactory.pxi
index 233654e9..3973554c 100644
--- a/python/src/sa/rulefactory.pxi
+++ b/python/src/sa/rulefactory.pxi
@@ -1101,9 +1101,8 @@ cdef class HieroCachingRuleFactory:
                             for f, elist in fphrases.iteritems():
                                 for e, alslist in elist.iteritems():
                                     alignment, max_locs = max(alslist.iteritems(), key=lambda x: len(x[1]))
-                                    locs = tuple(itertools.chain(alslist.itervalues()))
-                                    # count = len(locs) # Should be?
-                                    count = len(max_locs) # Was
+                                    locs = tuple(itertools.chain.from_iterable(alslist.itervalues()))
+                                    count = len(locs)
                                     scores = self.scorer.score(FeatureContext(
                                                f, e, count, fcount[f], num_samples,
                                                (k,i+spanlen), locs, fwords, self.fda, self.eda