[cdec.sa] Fix API to make everyone happy

3 files changed, 1495 insertions, 1250 deletions

diff --git a/python/src/sa/_sa.c b/python/src/sa/_sa.c
index d1cc55c0..5507cd15 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 14:37:00 2012 */
+/* Generated by Cython 0.17 on Thu Sep  6 16:40:54 2012 */
 
 #define PY_SSIZE_T_CLEAN
 #include "Python.h"
@@ -701,8 +701,8 @@ struct __pyx_obj_3_sa_LCP {
  * from libc.string cimport memset, strcpy
  * 
  * cdef class DataArray:             # <<<<<<<<<<<<<<
- *     cdef word2id
- *     cdef id2word
+ *     cdef public word2id
+ *     cdef public id2word
  */
 struct __pyx_obj_3_sa_DataArray {
   PyObject_HEAD
@@ -837,7 +837,7 @@ struct __pyx_obj_3_sa___pyx_scope_struct_6_genexpr {
 };
 
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":73
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":74
  *     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":71
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":72
  *             self.read_text_data(fp)
  * 
  *     def read_bitext(self, char* filename, int side):             # <<<<<<<<<<<<<<
@@ -1344,8 +1344,8 @@ static struct __pyx_vtabstruct_3_sa_StringMap *__pyx_vtabptr_3_sa_StringMap;
  * from libc.string cimport memset, strcpy
  * 
  * cdef class DataArray:             # <<<<<<<<<<<<<<
- *     cdef word2id
- *     cdef id2word
+ *     cdef public word2id
+ *     cdef public id2word
  */
 
 struct __pyx_vtabstruct_3_sa_DataArray {
@@ -2306,9 +2306,9 @@ static int __pyx_pf_3_sa_9DataArray___cinit__(struct __pyx_obj_3_sa_DataArray *_
 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_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_8get_id(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_word); /* proto */
+static PyObject *__pyx_pf_3_sa_9DataArray_10__getitem__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_loc); /* proto */
+static PyObject *__pyx_pf_3_sa_9DataArray_12get_sentence_bounds(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_loc); /* 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 */
@@ -2318,9 +2318,21 @@ static PyObject *__pyx_pf_3_sa_9DataArray_22read_binary(struct __pyx_obj_3_sa_Da
 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_7word2id___get__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self); /* proto */
+static int __pyx_pf_3_sa_9DataArray_7word2id_2__set__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_value); /* proto */
+static int __pyx_pf_3_sa_9DataArray_7word2id_4__del__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self); /* proto */
+static PyObject *__pyx_pf_3_sa_9DataArray_7id2word___get__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self); /* proto */
+static int __pyx_pf_3_sa_9DataArray_7id2word_2__set__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_value); /* proto */
+static int __pyx_pf_3_sa_9DataArray_7id2word_4__del__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self); /* 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_9DataArray_7sent_id___get__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self); /* proto */
+static int __pyx_pf_3_sa_9DataArray_7sent_id_2__set__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_value); /* proto */
+static int __pyx_pf_3_sa_9DataArray_7sent_id_4__del__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self); /* proto */
+static PyObject *__pyx_pf_3_sa_9DataArray_10sent_index___get__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self); /* proto */
+static int __pyx_pf_3_sa_9DataArray_10sent_index_2__set__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_value); /* proto */
+static int __pyx_pf_3_sa_9DataArray_10sent_index_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 */
@@ -2413,16 +2425,13 @@ static PyObject *__pyx_pf_3_sa_14Precomputation_4write_binary(struct __pyx_obj_3
 static PyObject *__pyx_pf_3_sa_14Precomputation_6precompute(struct __pyx_obj_3_sa_Precomputation *__pyx_v_self, PyObject *__pyx_v_stats, struct __pyx_obj_3_sa_SuffixArray *__pyx_v_sarray); /* proto */
 static int __pyx_pf_3_sa_11SuffixArray___cinit__(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, PyObject *__pyx_v_from_binary, PyObject *__pyx_v_from_text, PyObject *__pyx_v_side); /* proto */
 static PyObject *__pyx_pf_3_sa_11SuffixArray_2__getitem__(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, PyObject *__pyx_v_i); /* proto */
-static PyObject *__pyx_pf_3_sa_11SuffixArray_4get_sentence_id(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, PyObject *__pyx_v_i); /* proto */
-static PyObject *__pyx_pf_3_sa_11SuffixArray_6get_sentence(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, PyObject *__pyx_v_i); /* proto */
-static PyObject *__pyx_pf_3_sa_11SuffixArray_8get_sentence_position(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, PyObject *__pyx_v_loc); /* proto */
-static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, PyObject *__pyx_v_filename, PyObject *__pyx_v_side); /* proto */
-static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, int __pyx_v_i, int __pyx_v_j, int __pyx_v_h, struct __pyx_obj_3_sa_IntList *__pyx_v_isa, PyObject *__pyx_v_pad); /* proto */
-static PyObject *__pyx_pf_3_sa_11SuffixArray_14write_text(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, char *__pyx_v_filename); /* proto */
-static PyObject *__pyx_pf_3_sa_11SuffixArray_16read_binary(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, char *__pyx_v_filename); /* proto */
-static PyObject *__pyx_pf_3_sa_11SuffixArray_18write_binary(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, char *__pyx_v_filename); /* proto */
-static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, char *__pyx_v_filename); /* proto */
-static PyObject *__pyx_pf_3_sa_11SuffixArray_22lookup(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, PyObject *__pyx_v_word, int __pyx_v_offset, int __pyx_v_low, int __pyx_v_high); /* proto */
+static PyObject *__pyx_pf_3_sa_11SuffixArray_4read_text(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, PyObject *__pyx_v_filename, PyObject *__pyx_v_side); /* proto */
+static PyObject *__pyx_pf_3_sa_11SuffixArray_6q3sort(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, int __pyx_v_i, int __pyx_v_j, int __pyx_v_h, struct __pyx_obj_3_sa_IntList *__pyx_v_isa, PyObject *__pyx_v_pad); /* proto */
+static PyObject *__pyx_pf_3_sa_11SuffixArray_8write_text(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, char *__pyx_v_filename); /* proto */
+static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_binary(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, char *__pyx_v_filename); /* proto */
+static PyObject *__pyx_pf_3_sa_11SuffixArray_12write_binary(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, char *__pyx_v_filename); /* proto */
+static PyObject *__pyx_pf_3_sa_11SuffixArray_14write_enhanced(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, char *__pyx_v_filename); /* proto */
+static PyObject *__pyx_pf_3_sa_11SuffixArray_16lookup(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, PyObject *__pyx_v_word, int __pyx_v_offset, int __pyx_v_low, int __pyx_v_high); /* proto */
 static int __pyx_pf_3_sa_8TrieNode___cinit__(struct __pyx_obj_3_sa_TrieNode *__pyx_v_self); /* proto */
 static PyObject *__pyx_pf_3_sa_8TrieNode_8children___get__(struct __pyx_obj_3_sa_TrieNode *__pyx_v_self); /* proto */
 static int __pyx_pf_3_sa_8TrieNode_8children_2__set__(struct __pyx_obj_3_sa_TrieNode *__pyx_v_self, PyObject *__pyx_v_value); /* proto */
@@ -2515,13 +2524,12 @@ static char __pyx_k_85[] = "RANK %d\tCOUNT, COST: %d    %d\tCUMUL: %d, %d";
 static char __pyx_k_86[] = "Precomputed collocations for %d patterns out of %d possible (upper bound %d)";
 static char __pyx_k_87[] = "Precomputed inverted index for %d patterns ";
 static char __pyx_k_88[] = "Precomputation took %f seconds";
-static char __pyx_k_89[] = "get_sentence_position";
-static char __pyx_k_90[] = "    Bucket sort took %f seconds";
-static char __pyx_k_91[] = "    Refining, sort depth = %d";
-static char __pyx_k_92[] = "    Refinement took %f seconds";
-static char __pyx_k_93[] = "    Finalizing sort...";
-static char __pyx_k_95[] = "Suffix array construction took %f seconds";
-static char __pyx_k_96[] = "Unexpected condition found in q3sort: sort from %d to %d";
+static char __pyx_k_89[] = "    Bucket sort took %f seconds";
+static char __pyx_k_90[] = "    Refining, sort depth = %d";
+static char __pyx_k_91[] = "    Refinement took %f seconds";
+static char __pyx_k_92[] = "    Finalizing sort...";
+static char __pyx_k_94[] = "Suffix array construction took %f seconds";
+static char __pyx_k_95[] = "Unexpected condition found in q3sort: sort from %d to %d";
 static char __pyx_k__0[] = "0";
 static char __pyx_k__1[] = "1";
 static char __pyx_k__e[] = "e";
@@ -2531,42 +2539,42 @@ static char __pyx_k__i[] = "i";
 static char __pyx_k__j[] = "j";
 static char __pyx_k__r[] = "r";
 static char __pyx_k__w[] = "w";
-static char __pyx_k_101[] = "Sampling strategy: uniform, max sample size = %d";
-static char __pyx_k_102[] = "Sampling strategy: no sampling";
-static char __pyx_k_104[] = "require_aligned_terminal";
-static char __pyx_k_105[] = "require_aligned_chunks";
-static char __pyx_k_106[] = "[X]";
-static char __pyx_k_107[] = "Must specify an alignment object";
-static char __pyx_k_109[] = "Reading precomputed data from file %s... ";
-static char __pyx_k_110[] = "Precomputation done with max nonterminals %d, decoder uses %d";
-static char __pyx_k_111[] = "Precomputation done with max terminals %d, decoder uses %d";
-static char __pyx_k_112[] = "Precomputation done with max initial size %d, decoder uses %d";
-static char __pyx_k_113[] = "Precomputation done with min gap size %d, decoder uses %d";
-static char __pyx_k_114[] = "Converting %d hash keys on precomputed inverted index... ";
-static char __pyx_k_115[] = "Converting %d hash keys on precomputed collocations... ";
-static char __pyx_k_116[] = "Processing precomputations took %f seconds";
-static char __pyx_k_117[] = "{";
-  static char __pyx_k_118[] = "(";
-static char __pyx_k_119[] = "}";
-static char __pyx_k_120[] = "get_precomputed_collocation";
-static char __pyx_k_121[] = "double binary";
-static char __pyx_k_122[] = "Keyword trie error";
-static char __pyx_k_124[] = "get_all_nodes_isteps_away";
-static char __pyx_k_125[] = "Total time for rule lookup, extraction, and scoring = %f seconds";
-static char __pyx_k_126[] = "    Extract time = %f seconds";
-static char __pyx_k_127[] = "No aligned terminals";
-static char __pyx_k_128[] = "Unaligned chunk";
-static char __pyx_k_129[] = "Gaps are not tight phrases";
-static char __pyx_k_130[] = "Inside edges of preceding subphrase are not tight";
-static char __pyx_k_131[] = "Inside edges of following subphrase are not tight";
-static char __pyx_k_132[] = "Subphrase [%d, %d] failed integrity check";
-static char __pyx_k_133[] = "Didn't extract anything from [%d, %d] -> [%d, %d]";
-static char __pyx_k_134[] = "Unable to extract basic phrase";
-static char __pyx_k_135[] = "%s=%s";
-static char __pyx_k_138[] = "/Users/vchahun/Sandbox/cdec/python/src/sa/_sa.pyx";
-static char __pyx_k_139[] = "cdec.sa";
-static char __pyx_k_143[] = "/Users/vchahun/Sandbox/cdec/python/src/sa/sym.pxi";
-static char __pyx_k_148[] = "*EPS*";
+static char __pyx_k_100[] = "Sampling strategy: uniform, max sample size = %d";
+static char __pyx_k_101[] = "Sampling strategy: no sampling";
+static char __pyx_k_103[] = "require_aligned_terminal";
+static char __pyx_k_104[] = "require_aligned_chunks";
+static char __pyx_k_105[] = "[X]";
+static char __pyx_k_106[] = "Must specify an alignment object";
+static char __pyx_k_108[] = "Reading precomputed data from file %s... ";
+static char __pyx_k_109[] = "Precomputation done with max nonterminals %d, decoder uses %d";
+static char __pyx_k_110[] = "Precomputation done with max terminals %d, decoder uses %d";
+static char __pyx_k_111[] = "Precomputation done with max initial size %d, decoder uses %d";
+static char __pyx_k_112[] = "Precomputation done with min gap size %d, decoder uses %d";
+static char __pyx_k_113[] = "Converting %d hash keys on precomputed inverted index... ";
+static char __pyx_k_114[] = "Converting %d hash keys on precomputed collocations... ";
+static char __pyx_k_115[] = "Processing precomputations took %f seconds";
+static char __pyx_k_116[] = "{";
+  static char __pyx_k_117[] = "(";
+static char __pyx_k_118[] = "}";
+static char __pyx_k_119[] = "get_precomputed_collocation";
+static char __pyx_k_120[] = "double binary";
+static char __pyx_k_121[] = "Keyword trie error";
+static char __pyx_k_123[] = "get_all_nodes_isteps_away";
+static char __pyx_k_124[] = "Total time for rule lookup, extraction, and scoring = %f seconds";
+static char __pyx_k_125[] = "    Extract time = %f seconds";
+static char __pyx_k_126[] = "No aligned terminals";
+static char __pyx_k_127[] = "Unaligned chunk";
+static char __pyx_k_128[] = "Gaps are not tight phrases";
+static char __pyx_k_129[] = "Inside edges of preceding subphrase are not tight";
+static char __pyx_k_130[] = "Inside edges of following subphrase are not tight";
+static char __pyx_k_131[] = "Subphrase [%d, %d] failed integrity check";
+static char __pyx_k_132[] = "Didn't extract anything from [%d, %d] -> [%d, %d]";
+static char __pyx_k_133[] = "Unable to extract basic phrase";
+static char __pyx_k_134[] = "%s=%s";
+static char __pyx_k_137[] = "/Users/vchahun/Sandbox/cdec/python/src/sa/_sa.pyx";
+static char __pyx_k_138[] = "cdec.sa";
+static char __pyx_k_142[] = "/Users/vchahun/Sandbox/cdec/python/src/sa/sym.pxi";
+static char __pyx_k_147[] = "*EPS*";
 static char __pyx_k__gc[] = "gc";
 static char __pyx_k__sa[] = "sa";
 static char __pyx_k___sa[] = "_sa";
@@ -2721,7 +2729,6 @@ static char __pyx_k__sample_size[] = "sample_size";
 static char __pyx_k__suffix_link[] = "suffix_link";
 static char __pyx_k__use_sent_id[] = "use_sent_id";
 static char __pyx_k___doquicksort[] = "_doquicksort";
-static char __pyx_k__get_sentence[] = "get_sentence";
 static char __pyx_k__gzip_or_text[] = "gzip_or_text";
 static char __pyx_k__make_lattice[] = "make_lattice";
 static char __pyx_k__min_gap_size[] = "min_gap_size";
@@ -2738,7 +2745,6 @@ static char __pyx_k__read_text_data[] = "read_text_data";
 static char __pyx_k__by_slack_factor[] = "by_slack_factor";
 static char __pyx_k__decode_sentence[] = "decode_sentence";
 static char __pyx_k__get_next_states[] = "get_next_states";
-static char __pyx_k__get_sentence_id[] = "get_sentence_id";
 static char __pyx_k__num_subpatterns[] = "num_subpatterns";
 static char __pyx_k__phrase_location[] = "phrase_location";
 static char __pyx_k__precompute_file[] = "precompute_file";
@@ -2755,11 +2761,12 @@ static char __pyx_k__max_target_length[] = "max_target_length";
 static char __pyx_k__train_min_gap_size[] = "train_min_gap_size";
 static char __pyx_k__pattern2phrase_plus[] = "pattern2phrase_plus";
 static PyObject *__pyx_kp_s_1;
+static PyObject *__pyx_kp_s_100;
 static PyObject *__pyx_kp_s_101;
-static PyObject *__pyx_kp_s_102;
+static PyObject *__pyx_n_s_103;
 static PyObject *__pyx_n_s_104;
-static PyObject *__pyx_n_s_105;
-static PyObject *__pyx_kp_s_107;
+static PyObject *__pyx_kp_s_106;
+static PyObject *__pyx_kp_s_108;
 static PyObject *__pyx_kp_s_109;
 static PyObject *__pyx_kp_s_110;
 static PyObject *__pyx_kp_s_111;
@@ -2770,11 +2777,11 @@ static PyObject *__pyx_kp_s_115;
 static PyObject *__pyx_kp_s_116;
 static PyObject *__pyx_kp_s_117;
 static PyObject *__pyx_kp_s_118;
-static PyObject *__pyx_kp_s_119;
-static PyObject *__pyx_n_s_120;
+static PyObject *__pyx_n_s_119;
+static PyObject *__pyx_kp_s_120;
 static PyObject *__pyx_kp_s_121;
-static PyObject *__pyx_kp_s_122;
-static PyObject *__pyx_n_s_124;
+static PyObject *__pyx_n_s_123;
+static PyObject *__pyx_kp_s_124;
 static PyObject *__pyx_kp_s_125;
 static PyObject *__pyx_kp_s_126;
 static PyObject *__pyx_kp_s_127;
@@ -2786,11 +2793,10 @@ static PyObject *__pyx_kp_s_131;
 static PyObject *__pyx_kp_s_132;
 static PyObject *__pyx_kp_s_133;
 static PyObject *__pyx_kp_s_134;
-static PyObject *__pyx_kp_s_135;
+static PyObject *__pyx_kp_s_137;
 static PyObject *__pyx_kp_s_138;
-static PyObject *__pyx_kp_s_139;
 static PyObject *__pyx_kp_s_14;
-static PyObject *__pyx_kp_s_143;
+static PyObject *__pyx_kp_s_142;
 static PyObject *__pyx_kp_s_18;
 static PyObject *__pyx_kp_s_2;
 static PyObject *__pyx_kp_s_21;
@@ -2833,14 +2839,13 @@ static PyObject *__pyx_kp_s_85;
 static PyObject *__pyx_kp_s_86;
 static PyObject *__pyx_kp_s_87;
 static PyObject *__pyx_kp_s_88;
-static PyObject *__pyx_n_s_89;
+static PyObject *__pyx_kp_s_89;
 static PyObject *__pyx_kp_s_9;
 static PyObject *__pyx_kp_s_90;
 static PyObject *__pyx_kp_s_91;
 static PyObject *__pyx_kp_s_92;
-static PyObject *__pyx_kp_s_93;
+static PyObject *__pyx_kp_s_94;
 static PyObject *__pyx_kp_s_95;
-static PyObject *__pyx_kp_s_96;
 static PyObject *__pyx_kp_s__0;
 static PyObject *__pyx_kp_s__1;
 static PyObject *__pyx_n_s__Counter;
@@ -2918,8 +2923,6 @@ static PyObject *__pyx_n_s__get_e_id;
 static PyObject *__pyx_n_s__get_f_id;
 static PyObject *__pyx_n_s__get_id;
 static PyObject *__pyx_n_s__get_next_states;
-static PyObject *__pyx_n_s__get_sentence;
-static PyObject *__pyx_n_s__get_sentence_id;
 static PyObject *__pyx_n_s__get_word;
 static PyObject *__pyx_n_s__getchunk;
 static PyObject *__pyx_n_s__getrusage;
@@ -3046,7 +3049,7 @@ static PyObject *__pyx_int_20;
 static PyObject *__pyx_int_1000;
 static PyObject *__pyx_int_65536;
 static PyObject *__pyx_k_41;
-static PyObject *__pyx_k_100;
+static PyObject *__pyx_k_99;
 static PyObject *__pyx_k_tuple_10;
 static PyObject *__pyx_k_tuple_11;
 static PyObject *__pyx_k_tuple_12;
@@ -3088,22 +3091,22 @@ static PyObject *__pyx_k_tuple_79;
 static PyObject *__pyx_k_tuple_80;
 static PyObject *__pyx_k_tuple_81;
 static PyObject *__pyx_k_tuple_82;
-static PyObject *__pyx_k_tuple_94;
+static PyObject *__pyx_k_tuple_93;
+static PyObject *__pyx_k_tuple_96;
 static PyObject *__pyx_k_tuple_97;
 static PyObject *__pyx_k_tuple_98;
-static PyObject *__pyx_k_tuple_99;
-static PyObject *__pyx_k_tuple_103;
-static PyObject *__pyx_k_tuple_108;
-static PyObject *__pyx_k_tuple_123;
-static PyObject *__pyx_k_tuple_136;
+static PyObject *__pyx_k_tuple_102;
+static PyObject *__pyx_k_tuple_107;
+static PyObject *__pyx_k_tuple_122;
+static PyObject *__pyx_k_tuple_135;
+static PyObject *__pyx_k_tuple_139;
 static PyObject *__pyx_k_tuple_140;
-static PyObject *__pyx_k_tuple_141;
-static PyObject *__pyx_k_tuple_144;
-static PyObject *__pyx_k_tuple_146;
-static PyObject *__pyx_k_codeobj_137;
-static PyObject *__pyx_k_codeobj_142;
-static PyObject *__pyx_k_codeobj_145;
-static PyObject *__pyx_k_codeobj_147;
+static PyObject *__pyx_k_tuple_143;
+static PyObject *__pyx_k_tuple_145;
+static PyObject *__pyx_k_codeobj_136;
+static PyObject *__pyx_k_codeobj_141;
+static PyObject *__pyx_k_codeobj_144;
+static PyObject *__pyx_k_codeobj_146;
 
 /* "_sa.pyx":5
  * import gzip
@@ -7384,7 +7387,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_6get_sentence(struct __pyx_obj_3_sa_Da
  *             sent.append(self.id2word[self.data.arr[i]])
  *         return sent             # <<<<<<<<<<<<<<
  * 
- *     def get_sentence_position(self, loc):
+ *     def get_id(self, word):
  */
   __Pyx_XDECREF(__pyx_r);
   __Pyx_INCREF(((PyObject *)__pyx_v_sent));
@@ -7406,12 +7409,12 @@ static PyObject *__pyx_pf_3_sa_9DataArray_6get_sentence(struct __pyx_obj_3_sa_Da
 }
 
 /* Python wrapper */
-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) {
+static PyObject *__pyx_pw_3_sa_9DataArray_9get_id(PyObject *__pyx_v_self, PyObject *__pyx_v_word); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_9get_id(PyObject *__pyx_v_self, PyObject *__pyx_v_word) {
   PyObject *__pyx_r = 0;
   __Pyx_RefNannyDeclarations
-  __Pyx_RefNannySetupContext("get_sentence_position (wrapper)", 0);
-  __pyx_r = __pyx_pf_3_sa_9DataArray_8get_sentence_position(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((PyObject *)__pyx_v_loc));
+  __Pyx_RefNannySetupContext("get_id (wrapper)", 0);
+  __pyx_r = __pyx_pf_3_sa_9DataArray_8get_id(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((PyObject *)__pyx_v_word));
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
@@ -7419,73 +7422,12 @@ static PyObject *__pyx_pw_3_sa_9DataArray_9get_sentence_position(PyObject *__pyx
 /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":47
  *         return sent
  * 
- *     def get_sentence_position(self, loc):             # <<<<<<<<<<<<<<
- *         return loc - self.sent_index.arr[self.sent_id.arr[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;
-  PyObject *__pyx_t_2 = NULL;
-  PyObject *__pyx_t_3 = NULL;
-  int __pyx_lineno = 0;
-  const char *__pyx_filename = NULL;
-  int __pyx_clineno = 0;
-  __Pyx_RefNannySetupContext("get_sentence_position", 0);
-
-  /* "/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]]             # <<<<<<<<<<<<<<
- * 
- *     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 = 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 = 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;
-  __pyx_t_3 = 0;
-  goto __pyx_L0;
-
-  __pyx_r = Py_None; __Pyx_INCREF(Py_None);
-  goto __pyx_L0;
-  __pyx_L1_error:;
-  __Pyx_XDECREF(__pyx_t_2);
-  __Pyx_XDECREF(__pyx_t_3);
-  __Pyx_AddTraceback("_sa.DataArray.get_sentence_position", __pyx_clineno, __pyx_lineno, __pyx_filename);
-  __pyx_r = NULL;
-  __pyx_L0:;
-  __Pyx_XGIVEREF(__pyx_r);
-  __Pyx_RefNannyFinishContext();
-  return __pyx_r;
-}
-
-/* Python wrapper */
-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_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":50
- *         return loc - self.sent_index.arr[self.sent_id.arr[loc]]
- * 
  *     def get_id(self, word):             # <<<<<<<<<<<<<<
  *         if not word in self.word2id:
  *             self.word2id[word] = len(self.id2word)
  */
 
-static PyObject *__pyx_pf_3_sa_9DataArray_10get_id(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_word) {
+static PyObject *__pyx_pf_3_sa_9DataArray_8get_id(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_word) {
   PyObject *__pyx_r = NULL;
   __Pyx_RefNannyDeclarations
   int __pyx_t_1;
@@ -7497,18 +7439,18 @@ static PyObject *__pyx_pf_3_sa_9DataArray_10get_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":51
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":48
  * 
  *     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 = 51; __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 = 48; __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":52
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":49
  *     def get_id(self, word):
  *         if not word in self.word2id:
  *             self.word2id[word] = len(self.id2word)             # <<<<<<<<<<<<<<
@@ -7517,36 +7459,36 @@ static PyObject *__pyx_pf_3_sa_9DataArray_10get_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 = 52; __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 = 49; __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 = 52; __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 = 49; __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 = 52; __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 = 49; __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":53
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":50
  *         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 = 53; __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 = 50; __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":54
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":51
  *             self.word2id[word] = len(self.id2word)
  *             self.id2word.append(word)
  *         return self.word2id[word]             # <<<<<<<<<<<<<<
  * 
- *     def get_word(self, id):
+ *     def __getitem__(self, loc):
  */
   __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 = 54; __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 = 51; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __pyx_r = __pyx_t_3;
   __pyx_t_3 = 0;
@@ -7565,52 +7507,134 @@ static PyObject *__pyx_pf_3_sa_9DataArray_10get_id(struct __pyx_obj_3_sa_DataArr
 }
 
 /* Python wrapper */
-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) {
+static PyObject *__pyx_pw_3_sa_9DataArray_11__getitem__(PyObject *__pyx_v_self, PyObject *__pyx_v_loc); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_11__getitem__(PyObject *__pyx_v_self, PyObject *__pyx_v_loc) {
   PyObject *__pyx_r = 0;
   __Pyx_RefNannyDeclarations
-  __Pyx_RefNannySetupContext("get_word (wrapper)", 0);
-  __pyx_r = __pyx_pf_3_sa_9DataArray_12get_word(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self), ((PyObject *)__pyx_v_id));
+  __Pyx_RefNannySetupContext("__getitem__ (wrapper)", 0);
+  __pyx_r = __pyx_pf_3_sa_9DataArray_10__getitem__(((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":56
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":53
  *         return self.word2id[word]
  * 
- *     def get_word(self, id):             # <<<<<<<<<<<<<<
- *         return self.id2word[id]
+ *     def __getitem__(self, loc):             # <<<<<<<<<<<<<<
+ *         return self.id2word[self.data.arr[loc]]
  * 
  */
 
-static PyObject *__pyx_pf_3_sa_9DataArray_12get_word(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_id) {
+static PyObject *__pyx_pf_3_sa_9DataArray_10__getitem__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_loc) {
   PyObject *__pyx_r = NULL;
   __Pyx_RefNannyDeclarations
-  PyObject *__pyx_t_1 = NULL;
+  Py_ssize_t __pyx_t_1;
+  PyObject *__pyx_t_2 = NULL;
   int __pyx_lineno = 0;
   const char *__pyx_filename = NULL;
   int __pyx_clineno = 0;
-  __Pyx_RefNannySetupContext("get_word", 0);
+  __Pyx_RefNannySetupContext("__getitem__", 0);
+
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":54
+ * 
+ *     def __getitem__(self, loc):
+ *         return self.id2word[self.data.arr[loc]]             # <<<<<<<<<<<<<<
+ * 
+ *     def get_sentence_bounds(self, loc):
+ */
+  __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 = 54; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetItemInt(__pyx_v_self->id2word, (__pyx_v_self->data->arr[__pyx_t_1]), sizeof(int), PyInt_FromLong); if (!__pyx_t_2) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 54; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_2);
+  __pyx_r = __pyx_t_2;
+  __pyx_t_2 = 0;
+  goto __pyx_L0;
+
+  __pyx_r = Py_None; __Pyx_INCREF(Py_None);
+  goto __pyx_L0;
+  __pyx_L1_error:;
+  __Pyx_XDECREF(__pyx_t_2);
+  __Pyx_AddTraceback("_sa.DataArray.__getitem__", __pyx_clineno, __pyx_lineno, __pyx_filename);
+  __pyx_r = NULL;
+  __pyx_L0:;
+  __Pyx_XGIVEREF(__pyx_r);
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+/* Python wrapper */
+static PyObject *__pyx_pw_3_sa_9DataArray_13get_sentence_bounds(PyObject *__pyx_v_self, PyObject *__pyx_v_loc); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_13get_sentence_bounds(PyObject *__pyx_v_self, PyObject *__pyx_v_loc) {
+  PyObject *__pyx_r = 0;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("get_sentence_bounds (wrapper)", 0);
+  __pyx_r = __pyx_pf_3_sa_9DataArray_12get_sentence_bounds(((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":56
+ *         return self.id2word[self.data.arr[loc]]
+ * 
+ *     def get_sentence_bounds(self, loc):             # <<<<<<<<<<<<<<
+ *         cdef int sid = self.sent_id.arr[loc]
+ *         return (self.sent_index.arr[sid], self.sent_index.arr[sid+1])
+ */
+
+static PyObject *__pyx_pf_3_sa_9DataArray_12get_sentence_bounds(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_loc) {
+  int __pyx_v_sid;
+  PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
+  Py_ssize_t __pyx_t_1;
+  PyObject *__pyx_t_2 = NULL;
+  PyObject *__pyx_t_3 = NULL;
+  PyObject *__pyx_t_4 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
+  __Pyx_RefNannySetupContext("get_sentence_bounds", 0);
 
   /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":57
  * 
- *     def get_word(self, id):
- *         return self.id2word[id]             # <<<<<<<<<<<<<<
+ *     def get_sentence_bounds(self, loc):
+ *         cdef int sid = self.sent_id.arr[loc]             # <<<<<<<<<<<<<<
+ *         return (self.sent_index.arr[sid], self.sent_index.arr[sid+1])
+ * 
+ */
+  __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 = 57; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_v_sid = (__pyx_v_self->sent_id->arr[__pyx_t_1]);
+
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":58
+ *     def get_sentence_bounds(self, loc):
+ *         cdef int sid = self.sent_id.arr[loc]
+ *         return (self.sent_index.arr[sid], self.sent_index.arr[sid+1])             # <<<<<<<<<<<<<<
  * 
  *     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 = 57; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_r = __pyx_t_1;
-  __pyx_t_1 = 0;
+  __pyx_t_2 = PyInt_FromLong((__pyx_v_self->sent_index->arr[__pyx_v_sid])); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 58; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_2);
+  __pyx_t_3 = PyInt_FromLong((__pyx_v_self->sent_index->arr[(__pyx_v_sid + 1)])); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 58; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_3);
+  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 58; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
+  PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_2);
+  __Pyx_GIVEREF(__pyx_t_2);
+  PyTuple_SET_ITEM(__pyx_t_4, 1, __pyx_t_3);
+  __Pyx_GIVEREF(__pyx_t_3);
+  __pyx_t_2 = 0;
+  __pyx_t_3 = 0;
+  __pyx_r = ((PyObject *)__pyx_t_4);
+  __pyx_t_4 = 0;
   goto __pyx_L0;
 
   __pyx_r = Py_None; __Pyx_INCREF(Py_None);
   goto __pyx_L0;
   __pyx_L1_error:;
-  __Pyx_XDECREF(__pyx_t_1);
-  __Pyx_AddTraceback("_sa.DataArray.get_word", __pyx_clineno, __pyx_lineno, __pyx_filename);
+  __Pyx_XDECREF(__pyx_t_2);
+  __Pyx_XDECREF(__pyx_t_3);
+  __Pyx_XDECREF(__pyx_t_4);
+  __Pyx_AddTraceback("_sa.DataArray.get_sentence_bounds", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
   __Pyx_XGIVEREF(__pyx_r);
@@ -7626,7 +7650,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_15write_text(PyObject *__pyx_v_self, P
   __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 = 59; __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 = 60; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L3_error:;
@@ -7639,8 +7663,8 @@ static PyObject *__pyx_pw_3_sa_9DataArray_15write_text(PyObject *__pyx_v_self, P
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":59
- *         return self.id2word[id]
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":60
+ *         return (self.sent_index.arr[sid], self.sent_index.arr[sid+1])
  * 
  *     def write_text(self, char* filename):             # <<<<<<<<<<<<<<
  *         with open(filename, "w") as f:
@@ -7671,7 +7695,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_14write_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":60
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":61
  * 
  *     def write_text(self, char* filename):
  *         with open(filename, "w") as f:             # <<<<<<<<<<<<<<
@@ -7679,9 +7703,9 @@ static PyObject *__pyx_pf_3_sa_9DataArray_14write_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 = 60; __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 = 61; __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 = 60; __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 = 61; __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));
@@ -7689,14 +7713,14 @@ static PyObject *__pyx_pf_3_sa_9DataArray_14write_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 = 60; __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 = 61; __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 = 60; __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 = 61; __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 = 60; __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 = 61; __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 = 60; __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 = 61; __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;
@@ -7711,7 +7735,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_14write_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":61
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":62
  *     def write_text(self, char* filename):
  *         with open(filename, "w") as f:
  *             for w_id in self.data:             # <<<<<<<<<<<<<<
@@ -7722,7 +7746,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_14write_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 = 61; __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 = 62; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
             __Pyx_GOTREF(__pyx_t_4);
             __pyx_t_9 = Py_TYPE(__pyx_t_4)->tp_iternext;
           }
@@ -7730,23 +7754,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_14write_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 = 61; __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 = 62; __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 = 61; __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 = 62; __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 = 61; __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 = 62; __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 = 61; __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 = 62; __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 = 61; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+                  else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 62; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
                 }
                 break;
               }
@@ -7756,47 +7780,47 @@ static PyObject *__pyx_pf_3_sa_9DataArray_14write_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":62
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":63
  *         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 = 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_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 = 63; __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 = 63; __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":63
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":64
  *             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 = 63; __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 = 64; __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 = 63; __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 = 64; __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 = 63; __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 = 64; __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 = 63; __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 = 64; __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 = 63; __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 = 64; __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 = 63; __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 = 64; __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 = 63; __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 = 64; __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;
@@ -7805,28 +7829,28 @@ static PyObject *__pyx_pf_3_sa_9DataArray_14write_text(struct __pyx_obj_3_sa_Dat
             }
             __pyx_L18:;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":64
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":65
  *                 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 = 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_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 = 65; __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 = 65; __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":65
+              /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":66
  *                     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 = 65; __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 = 66; __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 = 65; __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 = 66; __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;
@@ -7847,7 +7871,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_14write_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":60
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":61
  * 
  *     def write_text(self, char* filename):
  *         with open(filename, "w") as f:             # <<<<<<<<<<<<<<
@@ -7856,11 +7880,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_14write_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 = 60; __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 = 61; __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 = 60; __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 = 61; __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);
@@ -7873,11 +7897,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_14write_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 = 60; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (unlikely(!__pyx_t_13)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 61; __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 = 60; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (unlikely(__pyx_t_10 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 61; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __pyx_t_14 = (!__pyx_t_10);
           if (__pyx_t_14) {
             __Pyx_GIVEREF(__pyx_t_4);
@@ -7885,7 +7909,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_14write_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 = 60; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+            {__pyx_filename = __pyx_f[3]; __pyx_lineno = 61; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
             goto __pyx_L22;
           }
           __pyx_L22:;
@@ -7913,11 +7937,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_14write_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 = 60; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 61; __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 = 60; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        if (unlikely(__pyx_t_14 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 61; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       }
     }
     goto __pyx_L23;
@@ -7953,7 +7977,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_17read_text(PyObject *__pyx_v_self, Py
   __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 = 67; __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 = 68; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L3_error:;
@@ -7966,7 +7990,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_17read_text(PyObject *__pyx_v_self, Py
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":67
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":68
  *                     f.write("\n")
  * 
  *     def read_text(self, char* filename):             # <<<<<<<<<<<<<<
@@ -7994,7 +8018,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16read_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":68
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":69
  * 
  *     def read_text(self, char* filename):
  *         with gzip_or_text(filename) as fp:             # <<<<<<<<<<<<<<
@@ -8002,24 +8026,24 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16read_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 = 68; __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 = 69; __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 = 68; __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 = 69; __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 = 68; __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 = 69; __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 = 68; __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 = 69; __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 = 68; __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 = 69; __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 = 68; __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 = 69; __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 = 68; __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 = 69; __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;
@@ -8034,21 +8058,21 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16read_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":69
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":70
  *     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 = 69; __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 = 70; __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 = 69; __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 = 70; __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 = 69; __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 = 70; __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;
@@ -8063,7 +8087,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16read_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":68
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":69
  * 
  *     def read_text(self, char* filename):
  *         with gzip_or_text(filename) as fp:             # <<<<<<<<<<<<<<
@@ -8072,11 +8096,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16read_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 = 68; __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 = 69; __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 = 68; __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 = 69; __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);
@@ -8089,11 +8113,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16read_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 = 68; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 69; __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 = 68; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (unlikely(__pyx_t_9 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 69; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __pyx_t_11 = (!__pyx_t_9);
           if (__pyx_t_11) {
             __Pyx_GIVEREF(__pyx_t_3);
@@ -8101,7 +8125,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16read_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 = 68; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+            {__pyx_filename = __pyx_f[3]; __pyx_lineno = 69; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
             goto __pyx_L18;
           }
           __pyx_L18:;
@@ -8129,11 +8153,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_16read_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 = 68; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 69; __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 = 68; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        if (unlikely(__pyx_t_11 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 69; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       }
     }
     goto __pyx_L19;
@@ -8187,11 +8211,11 @@ static PyObject *__pyx_pw_3_sa_9DataArray_19read_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 = 71; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+          __Pyx_RaiseArgtupleInvalid("read_bitext", 1, 2, 2, 1); {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __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 = 71; __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 = 72; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
     } else if (PyTuple_GET_SIZE(__pyx_args) != 2) {
       goto __pyx_L5_argtuple_error;
@@ -8199,12 +8223,12 @@ static PyObject *__pyx_pw_3_sa_9DataArray_19read_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 = 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;}
+    __pyx_v_filename = PyBytes_AsString(values[0]); if (unlikely((!__pyx_v_filename) && PyErr_Occurred())) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 72; __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 = 72; __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 = 71; __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 = 72; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
   __Pyx_AddTraceback("_sa.DataArray.read_bitext", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
@@ -8216,7 +8240,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_19read_bitext(PyObject *__pyx_v_self,
 }
 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":73
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":74
  *     def read_bitext(self, char* filename, int side):
  *         with gzip_or_text(filename) as fp:
  *             data = (line.split(' ||| ')[side] for line in fp)             # <<<<<<<<<<<<<<
@@ -8242,7 +8266,7 @@ 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 = 73; __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 = 74; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_cur_scope);
     __Pyx_RefNannyFinishContext();
     return (PyObject *) gen;
@@ -8279,13 +8303,13 @@ 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 = 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 (unlikely(!__pyx_sent_value)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 74; __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 = 74; __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 = 73; __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 = 74; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
     __pyx_t_3 = Py_TYPE(__pyx_t_1)->tp_iternext;
   }
@@ -8293,23 +8317,23 @@ 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 = 73; __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 = 74; __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 = 73; __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 = 74; __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 = 73; __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 = 74; __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 = 73; __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 = 74; __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 = 73; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 74; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         }
         break;
       }
@@ -8320,12 +8344,12 @@ 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 = 73; __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 = 74; __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 = 73; __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 = 74; __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 = 73; __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 = 74; __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;
@@ -8345,7 +8369,7 @@ 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 = 73; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    if (unlikely(!__pyx_sent_value)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 74; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   }
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
   PyErr_SetNone(PyExc_StopIteration);
@@ -8363,7 +8387,7 @@ static PyObject *__pyx_gb_3_sa_9DataArray_11read_bitext_2generator6(__pyx_Genera
   return NULL;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":71
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":72
  *             self.read_text_data(fp)
  * 
  *     def read_bitext(self, char* filename, int side):             # <<<<<<<<<<<<<<
@@ -8399,7 +8423,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_18read_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":72
+  /* "/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:             # <<<<<<<<<<<<<<
@@ -8407,24 +8431,24 @@ static PyObject *__pyx_pf_3_sa_9DataArray_18read_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 = 72; __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 = 73; __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 = 72; __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 = 73; __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 = 72; __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 = 73; __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 = 72; __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 = 73; __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 = 72; __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 = 73; __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 = 72; __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 = 73; __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 = 72; __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 = 73; __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;
@@ -8440,33 +8464,33 @@ static PyObject *__pyx_pf_3_sa_9DataArray_18read_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":73
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":74
  *     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 = 73; __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 = 74; __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":74
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":75
  *         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 = 74; __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 = 75; __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 = 74; __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 = 75; __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 = 74; __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 = 75; __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;
@@ -8481,7 +8505,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_18read_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":72
+        /* "/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:             # <<<<<<<<<<<<<<
@@ -8490,11 +8514,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_18read_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 = 72; __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 = 73; __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 = 72; __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 = 73; __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);
@@ -8507,11 +8531,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_18read_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 = 72; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __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 = 72; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (unlikely(__pyx_t_9 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __pyx_t_11 = (!__pyx_t_9);
           if (__pyx_t_11) {
             __Pyx_GIVEREF(__pyx_t_3);
@@ -8519,7 +8543,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_18read_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 = 72; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+            {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
             goto __pyx_L18;
           }
           __pyx_L18:;
@@ -8547,11 +8571,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_18read_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 = 72; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __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 = 72; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        if (unlikely(__pyx_t_11 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       }
     }
     goto __pyx_L19;
@@ -8589,7 +8613,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_21read_text_data(PyObject *__pyx_v_sel
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":76
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":77
  *             self.read_text_data(data)
  * 
  *     def read_text_data(self, data):             # <<<<<<<<<<<<<<
@@ -8619,7 +8643,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_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":77
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":78
  * 
  *     def read_text_data(self, data):
  *         cdef int word_count = 0             # <<<<<<<<<<<<<<
@@ -8628,7 +8652,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_text_data(struct __pyx_obj_3_sa
  */
   __pyx_v_word_count = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":78
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":79
  *     def read_text_data(self, data):
  *         cdef int word_count = 0
  *         for line_num, line in enumerate(data):             # <<<<<<<<<<<<<<
@@ -8641,7 +8665,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_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 = 78; __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 = 79; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_t_4 = Py_TYPE(__pyx_t_2)->tp_iternext;
   }
@@ -8649,23 +8673,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_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 = 78; __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 = 79; __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 = 78; __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 = 79; __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 = 78; __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 = 79; __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 = 78; __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 = 79; __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 = 78; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 79; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         }
         break;
       }
@@ -8677,43 +8701,43 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_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 = 78; __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 = 79; __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":79
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":80
  *         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 = 79; __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 = 80; __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 = 79; __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 = 80; __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":80
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":81
  *         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 = 80; __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 = 81; __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 = 80; __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 = 81; __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 = 80; __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 = 81; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_6);
       __pyx_t_8 = Py_TYPE(__pyx_t_6)->tp_iternext;
     }
@@ -8722,23 +8746,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_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 = 80; __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 = 81; __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 = 80; __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 = 81; __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 = 80; __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 = 81; __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 = 80; __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 = 81; __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 = 80; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 81; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
           }
           break;
         }
@@ -8748,30 +8772,30 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_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":81
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":82
  *             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 = 81; __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 = 82; __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 = 81; __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 = 82; __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 = 81; __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 = 82; __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 = 81; __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 = 82; __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":82
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":83
  *             for word in line.split():
  *                 self.data.append(self.get_id(word))
  *                 if self.use_sent_id:             # <<<<<<<<<<<<<<
@@ -8780,21 +8804,21 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_text_data(struct __pyx_obj_3_sa
  */
       if (__pyx_v_self->use_sent_id) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":83
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":84
  *                 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 = 83; __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 = 84; __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":84
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":85
  *                 if self.use_sent_id:
  *                     self.sent_id.append(line_num)
  *                 word_count = word_count + 1             # <<<<<<<<<<<<<<
@@ -8805,18 +8829,18 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_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":85
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":86
  *                     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 = 85; __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 = 86; __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":86
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":87
  *                 word_count = word_count + 1
  *             self.data.append(1)
  *             if self.use_sent_id:             # <<<<<<<<<<<<<<
@@ -8825,21 +8849,21 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_text_data(struct __pyx_obj_3_sa
  */
     if (__pyx_v_self->use_sent_id) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":87
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":88
  *             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 = 87; __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 = 88; __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":88
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":89
  *             if self.use_sent_id:
  *                 self.sent_id.append(line_num)
  *             word_count = word_count + 1             # <<<<<<<<<<<<<<
@@ -8851,27 +8875,27 @@ static PyObject *__pyx_pf_3_sa_9DataArray_20read_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":89
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":90
  *                 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 = 89; __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 = 90; __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":90
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":91
  *             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 = 90; __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 = 91; __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 = 90; __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 = 91; __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;
@@ -8904,7 +8928,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_23read_binary(PyObject *__pyx_v_self,
   __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 = 93; __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 = 94; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L3_error:;
@@ -8917,7 +8941,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_23read_binary(PyObject *__pyx_v_self,
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":93
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":94
  * 
  * 
  *     def read_binary(self, char* filename):             # <<<<<<<<<<<<<<
@@ -8931,7 +8955,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_22read_binary(struct __pyx_obj_3_sa_Da
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("read_binary", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":95
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":96
  *     def read_binary(self, char* filename):
  *         cdef FILE* f
  *         f = fopen(filename, "r")             # <<<<<<<<<<<<<<
@@ -8940,7 +8964,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_22read_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":96
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":97
  *         cdef FILE* f
  *         f = fopen(filename, "r")
  *         self.read_handle(f)             # <<<<<<<<<<<<<<
@@ -8949,7 +8973,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_22read_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":97
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":98
  *         f = fopen(filename, "r")
  *         self.read_handle(f)
  *         fclose(f)             # <<<<<<<<<<<<<<
@@ -8964,7 +8988,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_22read_binary(struct __pyx_obj_3_sa_Da
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":99
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":100
  *         fclose(f)
  * 
  *     cdef void read_handle(self, FILE* f):             # <<<<<<<<<<<<<<
@@ -8989,7 +9013,7 @@ 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":104
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":105
  *         cdef char* word
  * 
  *         self.data.read_handle(f)             # <<<<<<<<<<<<<<
@@ -8998,7 +9022,7 @@ 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":105
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":106
  * 
  *         self.data.read_handle(f)
  *         self.sent_index.read_handle(f)             # <<<<<<<<<<<<<<
@@ -9007,7 +9031,7 @@ 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":106
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":107
  *         self.data.read_handle(f)
  *         self.sent_index.read_handle(f)
  *         self.sent_id.read_handle(f)             # <<<<<<<<<<<<<<
@@ -9016,7 +9040,7 @@ 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":107
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":108
  *         self.sent_index.read_handle(f)
  *         self.sent_id.read_handle(f)
  *         fread(&(num_words), sizeof(int), 1, f)             # <<<<<<<<<<<<<<
@@ -9025,7 +9049,7 @@ 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":108
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":109
  *         self.sent_id.read_handle(f)
  *         fread(&(num_words), sizeof(int), 1, f)
  *         for i in range(num_words):             # <<<<<<<<<<<<<<
@@ -9036,7 +9060,7 @@ 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":109
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":110
  *         fread(&(num_words), sizeof(int), 1, f)
  *         for i in range(num_words):
  *             fread(&(word_len), sizeof(int), 1, f)             # <<<<<<<<<<<<<<
@@ -9045,7 +9069,7 @@ 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":110
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":111
  *         for i in range(num_words):
  *             fread(&(word_len), sizeof(int), 1, f)
  *             word = <char*> malloc (word_len * sizeof(char))             # <<<<<<<<<<<<<<
@@ -9054,7 +9078,7 @@ 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":111
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":112
  *             fread(&(word_len), sizeof(int), 1, f)
  *             word = <char*> malloc (word_len * sizeof(char))
  *             fread(word, sizeof(char), word_len, f)             # <<<<<<<<<<<<<<
@@ -9063,7 +9087,7 @@ 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":112
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":113
  *             word = <char*> malloc (word_len * sizeof(char))
  *             fread(word, sizeof(char), word_len, f)
  *             self.word2id[word] = len(self.id2word)             # <<<<<<<<<<<<<<
@@ -9072,31 +9096,31 @@ 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 = 112; __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 = 113; __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 = 112; __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 = 113; __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 = 112; __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 = 113; __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 = 112; __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 = 113; __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":113
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":114
  *             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 = 113; __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 = 114; __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 = 113; __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 = 114; __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":114
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":115
  *             self.word2id[word] = len(self.id2word)
  *             self.id2word.append(word)
  *             free(word)             # <<<<<<<<<<<<<<
@@ -9106,7 +9130,7 @@ 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":115
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":116
  *             self.id2word.append(word)
  *             free(word)
  *         if len(self.sent_id) == 0:             # <<<<<<<<<<<<<<
@@ -9115,12 +9139,12 @@ 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 = 115; __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 = 116; __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":116
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":117
  *             free(word)
  *         if len(self.sent_id) == 0:
  *             self.use_sent_id = False             # <<<<<<<<<<<<<<
@@ -9132,7 +9156,7 @@ 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":118
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":119
  *             self.use_sent_id = False
  *         else:
  *             self.use_sent_id = True             # <<<<<<<<<<<<<<
@@ -9152,7 +9176,7 @@ 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":120
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":121
  *             self.use_sent_id = True
  * 
  *     cdef void write_handle(self, FILE* f):             # <<<<<<<<<<<<<<
@@ -9176,7 +9200,7 @@ 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":124
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":125
  *         cdef int num_words
  * 
  *         self.data.write_handle(f)             # <<<<<<<<<<<<<<
@@ -9185,7 +9209,7 @@ 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":125
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":126
  * 
  *         self.data.write_handle(f)
  *         self.sent_index.write_handle(f)             # <<<<<<<<<<<<<<
@@ -9194,7 +9218,7 @@ 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":126
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":127
  *         self.data.write_handle(f)
  *         self.sent_index.write_handle(f)
  *         self.sent_id.write_handle(f)             # <<<<<<<<<<<<<<
@@ -9203,7 +9227,7 @@ 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":127
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":128
  *         self.sent_index.write_handle(f)
  *         self.sent_id.write_handle(f)
  *         num_words = len(self.id2word) - 2             # <<<<<<<<<<<<<<
@@ -9212,11 +9236,11 @@ 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 = 127; __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 = 128; __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":128
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":129
  *         self.sent_id.write_handle(f)
  *         num_words = len(self.id2word) - 2
  *         fwrite(&(num_words), sizeof(int), 1, f)             # <<<<<<<<<<<<<<
@@ -9225,20 +9249,20 @@ 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":129
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":130
  *         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 = 129; __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 = 130; __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 = 129; __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 = 130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
     __pyx_t_4 = Py_TYPE(__pyx_t_3)->tp_iternext;
   }
@@ -9247,23 +9271,23 @@ 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 = 129; __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 = 130; __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 = 129; __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 = 130; __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 = 129; __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 = 130; __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 = 129; __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 = 130; __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 = 129; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         }
         break;
       }
@@ -9273,17 +9297,17 @@ 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":130
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":131
  *         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 = 130; __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 = 131; __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":131
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":132
  *         for word in self.id2word[2:]:
  *             word_len = len(word) + 1
  *             fwrite(&(word_len), sizeof(int), 1, f)             # <<<<<<<<<<<<<<
@@ -9292,14 +9316,14 @@ 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":132
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":133
  *             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 = 132; __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 = 133; __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;
@@ -9322,7 +9346,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_25write_binary(PyObject *__pyx_v_self,
   __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 = 134; __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 = 135; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L3_error:;
@@ -9335,7 +9359,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_25write_binary(PyObject *__pyx_v_self,
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":134
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":135
  *             fwrite(<char *>word, sizeof(char), word_len, f)
  * 
  *     def write_binary(self, char* filename):             # <<<<<<<<<<<<<<
@@ -9349,7 +9373,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_24write_binary(struct __pyx_obj_3_sa_D
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("write_binary", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":136
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":137
  *     def write_binary(self, char* filename):
  *         cdef FILE* f
  *         f = fopen(filename, "w")             # <<<<<<<<<<<<<<
@@ -9358,7 +9382,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_24write_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":137
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":138
  *         cdef FILE* f
  *         f = fopen(filename, "w")
  *         self.write_handle(f)             # <<<<<<<<<<<<<<
@@ -9367,7 +9391,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_24write_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":138
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":139
  *         f = fopen(filename, "w")
  *         self.write_handle(f)
  *         fclose(f)             # <<<<<<<<<<<<<<
@@ -9393,7 +9417,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_27write_enhanced_handle(PyObject *__py
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":140
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":141
  *         fclose(f)
  * 
  *     def write_enhanced_handle(self, f):             # <<<<<<<<<<<<<<
@@ -9417,7 +9441,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_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":141
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":142
  * 
  *     def write_enhanced_handle(self, f):
  *         for i in self.data:             # <<<<<<<<<<<<<<
@@ -9428,7 +9452,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_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 = 141; __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 = 142; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
     __pyx_t_3 = Py_TYPE(__pyx_t_1)->tp_iternext;
   }
@@ -9436,23 +9460,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_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 = 141; __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 = 142; __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 = 141; __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 = 142; __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 = 141; __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 = 142; __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 = 141; __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 = 142; __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 = 141; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 142; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         }
         break;
       }
@@ -9462,23 +9486,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_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":142
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":143
  *     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 = 142; __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 = 143; __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 = 142; __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 = 143; __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 = 142; __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 = 143; __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 = 142; __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 = 143; __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;
@@ -9486,21 +9510,21 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_enhanced_handle(struct __pyx_o
   }
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":143
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":144
  *         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 = 143; __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 = 144; __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 = 143; __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 = 144; __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":144
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":145
  *             f.write("%d " %i)
  *         f.write("\n")
  *         for i in self.sent_index:             # <<<<<<<<<<<<<<
@@ -9511,7 +9535,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_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 = 144; __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 = 145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
     __pyx_t_3 = Py_TYPE(__pyx_t_5)->tp_iternext;
   }
@@ -9519,23 +9543,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_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 = 144; __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 = 145; __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 = 144; __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 = 145; __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 = 144; __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 = 145; __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 = 144; __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 = 145; __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 = 144; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         }
         break;
       }
@@ -9545,23 +9569,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_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":145
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":146
  *         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 = 145; __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 = 146; __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 = 145; __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 = 146; __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 = 145; __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 = 146; __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 = 145; __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 = 146; __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;
@@ -9569,21 +9593,21 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_enhanced_handle(struct __pyx_o
   }
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":146
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":147
  *         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 = 146; __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 = 147; __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 = 146; __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 = 147; __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":147
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":148
  *             f.write("%d " %i)
  *         f.write("\n")
  *         for i in self.sent_id:             # <<<<<<<<<<<<<<
@@ -9594,7 +9618,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_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 = 147; __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 = 148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_6);
     __pyx_t_3 = Py_TYPE(__pyx_t_6)->tp_iternext;
   }
@@ -9602,23 +9626,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_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 = 147; __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 = 148; __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 = 147; __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 = 148; __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 = 147; __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 = 148; __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 = 147; __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 = 148; __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 = 147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         }
         break;
       }
@@ -9628,23 +9652,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_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":148
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":149
  *         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 = 148; __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 = 149; __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 = 148; __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 = 149; __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 = 148; __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 = 149; __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 = 148; __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 = 149; __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;
@@ -9652,21 +9676,21 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_enhanced_handle(struct __pyx_o
   }
   __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":149
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":150
  *         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 = 149; __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 = 150; __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 = 149; __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 = 150; __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":150
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":151
  *             f.write("%d " %i)
  *         f.write("\n")
  *         for word in self.id2word:             # <<<<<<<<<<<<<<
@@ -9677,7 +9701,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_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 = 150; __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 = 151; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
     __pyx_t_3 = Py_TYPE(__pyx_t_4)->tp_iternext;
   }
@@ -9685,23 +9709,23 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_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 = 150; __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 = 151; __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 = 150; __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 = 151; __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 = 150; __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 = 151; __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 = 150; __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 = 151; __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 = 150; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          else {__pyx_filename = __pyx_f[3]; __pyx_lineno = 151; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         }
         break;
       }
@@ -9711,18 +9735,18 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_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":151
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":152
  *         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 = 151; __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 = 152; __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 = 151; __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 = 152; __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 = 151; __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 = 152; __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);
@@ -9730,15 +9754,15 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_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 = 151; __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 = 152; __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 = 151; __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 = 152; __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 = 151; __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 = 152; __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;
@@ -9746,16 +9770,16 @@ static PyObject *__pyx_pf_3_sa_9DataArray_26write_enhanced_handle(struct __pyx_o
   }
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":152
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":153
  *         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 = 152; __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 = 153; __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 = 152; __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 = 153; __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;
@@ -9785,7 +9809,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_29write_enhanced(PyObject *__pyx_v_sel
   __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 = 154; __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 = 155; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L3_error:;
@@ -9798,7 +9822,7 @@ static PyObject *__pyx_pw_3_sa_9DataArray_29write_enhanced(PyObject *__pyx_v_sel
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":154
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":155
  *         f.write("\n")
  * 
  *     def write_enhanced(self, char* filename):             # <<<<<<<<<<<<<<
@@ -9826,16 +9850,16 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_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":155
+  /* "/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)
  */
   /*with:*/ {
-    __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_t_1 = PyBytes_FromString(__pyx_v_filename); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 156; __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 = 155; __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 = 156; __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));
@@ -9843,14 +9867,14 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_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 = 155; __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 = 156; __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 = 155; __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 = 156; __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 = 155; __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 = 156; __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 = 155; __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 = 156; __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;
@@ -9865,14 +9889,14 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_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":156
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":157
  *     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 = 156; __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 = 157; __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 = 156; __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 = 157; __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));
@@ -9880,7 +9904,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_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 = 156; __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 = 157; __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;
@@ -9895,7 +9919,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_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":155
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":156
  * 
  *     def write_enhanced(self, char* filename):
  *         with open(filename, "w") as f:             # <<<<<<<<<<<<<<
@@ -9903,11 +9927,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_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 = 155; __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 = 156; __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 = 155; __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 = 156; __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);
@@ -9920,11 +9944,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_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 = 155; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 156; __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 = 155; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (unlikely(__pyx_t_9 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 156; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __pyx_t_11 = (!__pyx_t_9);
           if (__pyx_t_11) {
             __Pyx_GIVEREF(__pyx_t_2);
@@ -9932,7 +9956,7 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_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 = 155; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+            {__pyx_filename = __pyx_f[3]; __pyx_lineno = 156; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
             goto __pyx_L18;
           }
           __pyx_L18:;
@@ -9960,11 +9984,11 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_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 = 155; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 156; __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 = 155; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        if (unlikely(__pyx_t_11 < 0)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 156; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       }
     }
     goto __pyx_L19;
@@ -9991,6 +10015,180 @@ static PyObject *__pyx_pf_3_sa_9DataArray_28write_enhanced(struct __pyx_obj_3_sa
 }
 
 /* Python wrapper */
+static PyObject *__pyx_pw_3_sa_9DataArray_7word2id_1__get__(PyObject *__pyx_v_self); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_7word2id_1__get__(PyObject *__pyx_v_self) {
+  PyObject *__pyx_r = 0;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__get__ (wrapper)", 0);
+  __pyx_r = __pyx_pf_3_sa_9DataArray_7word2id___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":10
+ * 
+ * cdef class DataArray:
+ *     cdef public word2id             # <<<<<<<<<<<<<<
+ *     cdef public id2word
+ *     cdef public IntList data
+ */
+
+static PyObject *__pyx_pf_3_sa_9DataArray_7word2id___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(__pyx_v_self->word2id);
+  __pyx_r = __pyx_v_self->word2id;
+  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_7word2id_3__set__(PyObject *__pyx_v_self, PyObject *__pyx_v_value); /*proto*/
+static int __pyx_pw_3_sa_9DataArray_7word2id_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_7word2id_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_7word2id_2__set__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_value) {
+  int __pyx_r;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__set__", 0);
+  __Pyx_INCREF(__pyx_v_value);
+  __Pyx_GIVEREF(__pyx_v_value);
+  __Pyx_GOTREF(__pyx_v_self->word2id);
+  __Pyx_DECREF(__pyx_v_self->word2id);
+  __pyx_v_self->word2id = __pyx_v_value;
+
+  __pyx_r = 0;
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+/* Python wrapper */
+static int __pyx_pw_3_sa_9DataArray_7word2id_5__del__(PyObject *__pyx_v_self); /*proto*/
+static int __pyx_pw_3_sa_9DataArray_7word2id_5__del__(PyObject *__pyx_v_self) {
+  int __pyx_r;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__del__ (wrapper)", 0);
+  __pyx_r = __pyx_pf_3_sa_9DataArray_7word2id_4__del__(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self));
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+static int __pyx_pf_3_sa_9DataArray_7word2id_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->word2id);
+  __Pyx_DECREF(__pyx_v_self->word2id);
+  __pyx_v_self->word2id = Py_None;
+
+  __pyx_r = 0;
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+/* Python wrapper */
+static PyObject *__pyx_pw_3_sa_9DataArray_7id2word_1__get__(PyObject *__pyx_v_self); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_7id2word_1__get__(PyObject *__pyx_v_self) {
+  PyObject *__pyx_r = 0;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__get__ (wrapper)", 0);
+  __pyx_r = __pyx_pf_3_sa_9DataArray_7id2word___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":11
+ * cdef class DataArray:
+ *     cdef public word2id
+ *     cdef public id2word             # <<<<<<<<<<<<<<
+ *     cdef public IntList data
+ *     cdef public IntList sent_id
+ */
+
+static PyObject *__pyx_pf_3_sa_9DataArray_7id2word___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(__pyx_v_self->id2word);
+  __pyx_r = __pyx_v_self->id2word;
+  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_7id2word_3__set__(PyObject *__pyx_v_self, PyObject *__pyx_v_value); /*proto*/
+static int __pyx_pw_3_sa_9DataArray_7id2word_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_7id2word_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_7id2word_2__set__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self, PyObject *__pyx_v_value) {
+  int __pyx_r;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__set__", 0);
+  __Pyx_INCREF(__pyx_v_value);
+  __Pyx_GIVEREF(__pyx_v_value);
+  __Pyx_GOTREF(__pyx_v_self->id2word);
+  __Pyx_DECREF(__pyx_v_self->id2word);
+  __pyx_v_self->id2word = __pyx_v_value;
+
+  __pyx_r = 0;
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+/* Python wrapper */
+static int __pyx_pw_3_sa_9DataArray_7id2word_5__del__(PyObject *__pyx_v_self); /*proto*/
+static int __pyx_pw_3_sa_9DataArray_7id2word_5__del__(PyObject *__pyx_v_self) {
+  int __pyx_r;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__del__ (wrapper)", 0);
+  __pyx_r = __pyx_pf_3_sa_9DataArray_7id2word_4__del__(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self));
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+static int __pyx_pf_3_sa_9DataArray_7id2word_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->id2word);
+  __Pyx_DECREF(__pyx_v_self->id2word);
+  __pyx_v_self->id2word = Py_None;
+
+  __pyx_r = 0;
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+/* 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;
@@ -10002,11 +10200,11 @@ static PyObject *__pyx_pw_3_sa_9DataArray_4data_1__get__(PyObject *__pyx_v_self)
 }
 
 /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":12
- *     cdef word2id
- *     cdef id2word
+ *     cdef public word2id
+ *     cdef public id2word
  *     cdef public IntList data             # <<<<<<<<<<<<<<
- *     cdef IntList sent_id
- *     cdef IntList sent_index
+ *     cdef public IntList sent_id
+ *     cdef public IntList sent_index
  */
 
 static PyObject *__pyx_pf_3_sa_9DataArray_4data___get__(struct __pyx_obj_3_sa_DataArray *__pyx_v_self) {
@@ -10086,6 +10284,206 @@ static int __pyx_pf_3_sa_9DataArray_4data_4__del__(struct __pyx_obj_3_sa_DataArr
   return __pyx_r;
 }
 
+/* Python wrapper */
+static PyObject *__pyx_pw_3_sa_9DataArray_7sent_id_1__get__(PyObject *__pyx_v_self); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_7sent_id_1__get__(PyObject *__pyx_v_self) {
+  PyObject *__pyx_r = 0;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__get__ (wrapper)", 0);
+  __pyx_r = __pyx_pf_3_sa_9DataArray_7sent_id___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":13
+ *     cdef public id2word
+ *     cdef public IntList data
+ *     cdef public IntList sent_id             # <<<<<<<<<<<<<<
+ *     cdef public IntList sent_index
+ *     cdef bint use_sent_id
+ */
+
+static PyObject *__pyx_pf_3_sa_9DataArray_7sent_id___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->sent_id));
+  __pyx_r = ((PyObject *)__pyx_v_self->sent_id);
+  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_7sent_id_3__set__(PyObject *__pyx_v_self, PyObject *__pyx_v_value); /*proto*/
+static int __pyx_pw_3_sa_9DataArray_7sent_id_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_7sent_id_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_7sent_id_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 = 13; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_INCREF(__pyx_v_value);
+  __Pyx_GIVEREF(__pyx_v_value);
+  __Pyx_GOTREF(__pyx_v_self->sent_id);
+  __Pyx_DECREF(((PyObject *)__pyx_v_self->sent_id));
+  __pyx_v_self->sent_id = ((struct __pyx_obj_3_sa_IntList *)__pyx_v_value);
+
+  __pyx_r = 0;
+  goto __pyx_L0;
+  __pyx_L1_error:;
+  __Pyx_AddTraceback("_sa.DataArray.sent_id.__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_7sent_id_5__del__(PyObject *__pyx_v_self); /*proto*/
+static int __pyx_pw_3_sa_9DataArray_7sent_id_5__del__(PyObject *__pyx_v_self) {
+  int __pyx_r;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__del__ (wrapper)", 0);
+  __pyx_r = __pyx_pf_3_sa_9DataArray_7sent_id_4__del__(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self));
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+static int __pyx_pf_3_sa_9DataArray_7sent_id_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->sent_id);
+  __Pyx_DECREF(((PyObject *)__pyx_v_self->sent_id));
+  __pyx_v_self->sent_id = ((struct __pyx_obj_3_sa_IntList *)Py_None);
+
+  __pyx_r = 0;
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+/* Python wrapper */
+static PyObject *__pyx_pw_3_sa_9DataArray_10sent_index_1__get__(PyObject *__pyx_v_self); /*proto*/
+static PyObject *__pyx_pw_3_sa_9DataArray_10sent_index_1__get__(PyObject *__pyx_v_self) {
+  PyObject *__pyx_r = 0;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__get__ (wrapper)", 0);
+  __pyx_r = __pyx_pf_3_sa_9DataArray_10sent_index___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":14
+ *     cdef public IntList data
+ *     cdef public IntList sent_id
+ *     cdef public IntList sent_index             # <<<<<<<<<<<<<<
+ *     cdef bint use_sent_id
+ * 
+ */
+
+static PyObject *__pyx_pf_3_sa_9DataArray_10sent_index___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->sent_index));
+  __pyx_r = ((PyObject *)__pyx_v_self->sent_index);
+  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_10sent_index_3__set__(PyObject *__pyx_v_self, PyObject *__pyx_v_value); /*proto*/
+static int __pyx_pw_3_sa_9DataArray_10sent_index_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_10sent_index_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_10sent_index_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 = 14; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_INCREF(__pyx_v_value);
+  __Pyx_GIVEREF(__pyx_v_value);
+  __Pyx_GOTREF(__pyx_v_self->sent_index);
+  __Pyx_DECREF(((PyObject *)__pyx_v_self->sent_index));
+  __pyx_v_self->sent_index = ((struct __pyx_obj_3_sa_IntList *)__pyx_v_value);
+
+  __pyx_r = 0;
+  goto __pyx_L0;
+  __pyx_L1_error:;
+  __Pyx_AddTraceback("_sa.DataArray.sent_index.__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_10sent_index_5__del__(PyObject *__pyx_v_self); /*proto*/
+static int __pyx_pw_3_sa_9DataArray_10sent_index_5__del__(PyObject *__pyx_v_self) {
+  int __pyx_r;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__del__ (wrapper)", 0);
+  __pyx_r = __pyx_pf_3_sa_9DataArray_10sent_index_4__del__(((struct __pyx_obj_3_sa_DataArray *)__pyx_v_self));
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+static int __pyx_pf_3_sa_9DataArray_10sent_index_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->sent_index);
+  __Pyx_DECREF(((PyObject *)__pyx_v_self->sent_index));
+  __pyx_v_self->sent_index = ((struct __pyx_obj_3_sa_IntList *)Py_None);
+
+  __pyx_r = 0;
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/alignment.pxi":12
+ *     cdef IntList sent_index
+ * 
+ *     cdef int link(self, int i, int j):             # <<<<<<<<<<<<<<
+ *         """Integerizes an alignment link pair"""
+ *         return i*65536 + j
+ */
+
 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
@@ -32650,7 +33048,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_2__getitem__(struct __pyx_obj_3_sa_
  *     def __getitem__(self, i):
  *         return self.sa.arr[i]             # <<<<<<<<<<<<<<
  * 
- *     def get_sentence_id(self, i):
+ *     def read_text(self, filename, side):
  */
   __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[12]; __pyx_lineno = 21; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
@@ -32673,210 +33071,9 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_2__getitem__(struct __pyx_obj_3_sa_
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_11SuffixArray_5get_sentence_id(PyObject *__pyx_v_self, PyObject *__pyx_v_i); /*proto*/
-static PyObject *__pyx_pw_3_sa_11SuffixArray_5get_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_11SuffixArray_4get_sentence_id(((struct __pyx_obj_3_sa_SuffixArray *)__pyx_v_self), ((PyObject *)__pyx_v_i));
-  __Pyx_RefNannyFinishContext();
-  return __pyx_r;
-}
-
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":23
- *         return self.sa.arr[i]
- * 
- *     def get_sentence_id(self, i):             # <<<<<<<<<<<<<<
- *         return self.darray.get_sentence_id(i)
- * 
- */
-
-static PyObject *__pyx_pf_3_sa_11SuffixArray_4get_sentence_id(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, PyObject *__pyx_v_i) {
-  PyObject *__pyx_r = NULL;
-  __Pyx_RefNannyDeclarations
-  PyObject *__pyx_t_1 = NULL;
-  PyObject *__pyx_t_2 = NULL;
-  PyObject *__pyx_t_3 = NULL;
-  int __pyx_lineno = 0;
-  const char *__pyx_filename = NULL;
-  int __pyx_clineno = 0;
-  __Pyx_RefNannySetupContext("get_sentence_id", 0);
-
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":24
- * 
- *     def get_sentence_id(self, i):
- *         return self.darray.get_sentence_id(i)             # <<<<<<<<<<<<<<
- * 
- *     def get_sentence(self, i):
- */
-  __Pyx_XDECREF(__pyx_r);
-  __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self->darray), __pyx_n_s__get_sentence_id); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 24; __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[12]; __pyx_lineno = 24; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_2);
-  __Pyx_INCREF(__pyx_v_i);
-  PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_v_i);
-  __Pyx_GIVEREF(__pyx_v_i);
-  __pyx_t_3 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 24; __pyx_clineno = __LINE__; goto __pyx_L1_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;
-  __pyx_r = __pyx_t_3;
-  __pyx_t_3 = 0;
-  goto __pyx_L0;
-
-  __pyx_r = Py_None; __Pyx_INCREF(Py_None);
-  goto __pyx_L0;
-  __pyx_L1_error:;
-  __Pyx_XDECREF(__pyx_t_1);
-  __Pyx_XDECREF(__pyx_t_2);
-  __Pyx_XDECREF(__pyx_t_3);
-  __Pyx_AddTraceback("_sa.SuffixArray.get_sentence_id", __pyx_clineno, __pyx_lineno, __pyx_filename);
-  __pyx_r = NULL;
-  __pyx_L0:;
-  __Pyx_XGIVEREF(__pyx_r);
-  __Pyx_RefNannyFinishContext();
-  return __pyx_r;
-}
-
-/* Python wrapper */
-static PyObject *__pyx_pw_3_sa_11SuffixArray_7get_sentence(PyObject *__pyx_v_self, PyObject *__pyx_v_i); /*proto*/
-static PyObject *__pyx_pw_3_sa_11SuffixArray_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_11SuffixArray_6get_sentence(((struct __pyx_obj_3_sa_SuffixArray *)__pyx_v_self), ((PyObject *)__pyx_v_i));
-  __Pyx_RefNannyFinishContext();
-  return __pyx_r;
-}
-
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":26
- *         return self.darray.get_sentence_id(i)
- * 
- *     def get_sentence(self, i):             # <<<<<<<<<<<<<<
- *         return self.darray.get_sentence(i)
- * 
- */
-
-static PyObject *__pyx_pf_3_sa_11SuffixArray_6get_sentence(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, PyObject *__pyx_v_i) {
-  PyObject *__pyx_r = NULL;
-  __Pyx_RefNannyDeclarations
-  PyObject *__pyx_t_1 = NULL;
-  PyObject *__pyx_t_2 = NULL;
-  PyObject *__pyx_t_3 = NULL;
-  int __pyx_lineno = 0;
-  const char *__pyx_filename = NULL;
-  int __pyx_clineno = 0;
-  __Pyx_RefNannySetupContext("get_sentence", 0);
-
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":27
- * 
- *     def get_sentence(self, i):
- *         return self.darray.get_sentence(i)             # <<<<<<<<<<<<<<
- * 
- *     def get_sentence_position(self, loc):
- */
-  __Pyx_XDECREF(__pyx_r);
-  __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self->darray), __pyx_n_s__get_sentence); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 27; __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[12]; __pyx_lineno = 27; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_2);
-  __Pyx_INCREF(__pyx_v_i);
-  PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_v_i);
-  __Pyx_GIVEREF(__pyx_v_i);
-  __pyx_t_3 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 27; __pyx_clineno = __LINE__; goto __pyx_L1_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;
-  __pyx_r = __pyx_t_3;
-  __pyx_t_3 = 0;
-  goto __pyx_L0;
-
-  __pyx_r = Py_None; __Pyx_INCREF(Py_None);
-  goto __pyx_L0;
-  __pyx_L1_error:;
-  __Pyx_XDECREF(__pyx_t_1);
-  __Pyx_XDECREF(__pyx_t_2);
-  __Pyx_XDECREF(__pyx_t_3);
-  __Pyx_AddTraceback("_sa.SuffixArray.get_sentence", __pyx_clineno, __pyx_lineno, __pyx_filename);
-  __pyx_r = NULL;
-  __pyx_L0:;
-  __Pyx_XGIVEREF(__pyx_r);
-  __Pyx_RefNannyFinishContext();
-  return __pyx_r;
-}
-
-/* Python wrapper */
-static PyObject *__pyx_pw_3_sa_11SuffixArray_9get_sentence_position(PyObject *__pyx_v_self, PyObject *__pyx_v_loc); /*proto*/
-static PyObject *__pyx_pw_3_sa_11SuffixArray_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_11SuffixArray_8get_sentence_position(((struct __pyx_obj_3_sa_SuffixArray *)__pyx_v_self), ((PyObject *)__pyx_v_loc));
-  __Pyx_RefNannyFinishContext();
-  return __pyx_r;
-}
-
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":29
- *         return self.darray.get_sentence(i)
- * 
- *     def get_sentence_position(self, loc):             # <<<<<<<<<<<<<<
- *         return self.darray.get_sentence_position(loc)
- * 
- */
-
-static PyObject *__pyx_pf_3_sa_11SuffixArray_8get_sentence_position(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, PyObject *__pyx_v_loc) {
-  PyObject *__pyx_r = NULL;
-  __Pyx_RefNannyDeclarations
-  PyObject *__pyx_t_1 = NULL;
-  PyObject *__pyx_t_2 = NULL;
-  PyObject *__pyx_t_3 = NULL;
-  int __pyx_lineno = 0;
-  const char *__pyx_filename = NULL;
-  int __pyx_clineno = 0;
-  __Pyx_RefNannySetupContext("get_sentence_position", 0);
-
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":30
- * 
- *     def get_sentence_position(self, loc):
- *         return self.darray.get_sentence_position(loc)             # <<<<<<<<<<<<<<
- * 
- *     def read_text(self, filename, side):
- */
-  __Pyx_XDECREF(__pyx_r);
-  __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self->darray), __pyx_n_s_89); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 30; __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[12]; __pyx_lineno = 30; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_2);
-  __Pyx_INCREF(__pyx_v_loc);
-  PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_v_loc);
-  __Pyx_GIVEREF(__pyx_v_loc);
-  __pyx_t_3 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 30; __pyx_clineno = __LINE__; goto __pyx_L1_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;
-  __pyx_r = __pyx_t_3;
-  __pyx_t_3 = 0;
-  goto __pyx_L0;
-
-  __pyx_r = Py_None; __Pyx_INCREF(Py_None);
-  goto __pyx_L0;
-  __pyx_L1_error:;
-  __Pyx_XDECREF(__pyx_t_1);
-  __Pyx_XDECREF(__pyx_t_2);
-  __Pyx_XDECREF(__pyx_t_3);
-  __Pyx_AddTraceback("_sa.SuffixArray.get_sentence_position", __pyx_clineno, __pyx_lineno, __pyx_filename);
-  __pyx_r = NULL;
-  __pyx_L0:;
-  __Pyx_XGIVEREF(__pyx_r);
-  __Pyx_RefNannyFinishContext();
-  return __pyx_r;
-}
-
-/* Python wrapper */
-static PyObject *__pyx_pw_3_sa_11SuffixArray_11read_text(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_3_sa_11SuffixArray_10read_text[] = "Constructs suffix array using the algorithm\n        of Larsson & Sadahkane (1999)";
-static PyObject *__pyx_pw_3_sa_11SuffixArray_11read_text(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pw_3_sa_11SuffixArray_5read_text(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_3_sa_11SuffixArray_4read_text[] = "Constructs suffix array using the algorithm\n        of Larsson & Sadahkane (1999)";
+static PyObject *__pyx_pw_3_sa_11SuffixArray_5read_text(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_filename = 0;
   PyObject *__pyx_v_side = 0;
   PyObject *__pyx_r = 0;
@@ -32902,11 +33099,11 @@ static PyObject *__pyx_pw_3_sa_11SuffixArray_11read_text(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_text", 1, 2, 2, 1); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 32; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+          __Pyx_RaiseArgtupleInvalid("read_text", 1, 2, 2, 1); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 23; __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_text") < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 32; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+        if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "read_text") < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 23; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
     } else if (PyTuple_GET_SIZE(__pyx_args) != 2) {
       goto __pyx_L5_argtuple_error;
@@ -32919,26 +33116,26 @@ static PyObject *__pyx_pw_3_sa_11SuffixArray_11read_text(PyObject *__pyx_v_self,
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("read_text", 1, 2, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 32; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("read_text", 1, 2, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 23; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
   __Pyx_AddTraceback("_sa.SuffixArray.read_text", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_r = __pyx_pf_3_sa_11SuffixArray_10read_text(((struct __pyx_obj_3_sa_SuffixArray *)__pyx_v_self), __pyx_v_filename, __pyx_v_side);
+  __pyx_r = __pyx_pf_3_sa_11SuffixArray_4read_text(((struct __pyx_obj_3_sa_SuffixArray *)__pyx_v_self), __pyx_v_filename, __pyx_v_side);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":32
- *         return self.darray.get_sentence_position(loc)
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":23
+ *         return self.sa.arr[i]
  * 
  *     def read_text(self, filename, side):             # <<<<<<<<<<<<<<
  *         '''Constructs suffix array using the algorithm
  *         of Larsson & Sadahkane (1999)'''
  */
 
-static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, PyObject *__pyx_v_filename, PyObject *__pyx_v_side) {
+static PyObject *__pyx_pf_3_sa_11SuffixArray_4read_text(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, PyObject *__pyx_v_filename, PyObject *__pyx_v_side) {
   int __pyx_v_V;
   int __pyx_v_N;
   int __pyx_v_i;
@@ -32970,22 +33167,22 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("read_text", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":38
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":29
  *         cdef IntList isa, word_count
  * 
  *         self.darray = DataArray(from_text=filename, side=side, use_sent_id=True)             # <<<<<<<<<<<<<<
  *         N = len(self.darray)
  *         V = len(self.darray.id2word)
  */
-  __pyx_t_1 = PyDict_New(); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 38; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyDict_New(); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 29; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_1));
-  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_n_s__from_text), __pyx_v_filename) < 0) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 38; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_n_s__side), __pyx_v_side) < 0) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 38; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __pyx_t_2 = __Pyx_PyBool_FromLong(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 38; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_n_s__from_text), __pyx_v_filename) < 0) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 29; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_n_s__side), __pyx_v_side) < 0) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 29; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_PyBool_FromLong(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 29; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_n_s__use_sent_id), __pyx_t_2) < 0) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 38; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_n_s__use_sent_id), __pyx_t_2) < 0) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 29; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_DataArray)), ((PyObject *)__pyx_empty_tuple), ((PyObject *)__pyx_t_1)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 38; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_DataArray)), ((PyObject *)__pyx_empty_tuple), ((PyObject *)__pyx_t_1)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 29; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
   __Pyx_GIVEREF(__pyx_t_2);
@@ -32994,7 +33191,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
   __pyx_v_self->darray = ((struct __pyx_obj_3_sa_DataArray *)__pyx_t_2);
   __pyx_t_2 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":39
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":30
  * 
  *         self.darray = DataArray(from_text=filename, side=side, use_sent_id=True)
  *         N = len(self.darray)             # <<<<<<<<<<<<<<
@@ -33003,11 +33200,11 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
   __pyx_t_2 = ((PyObject *)__pyx_v_self->darray);
   __Pyx_INCREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_Length(__pyx_t_2); if (unlikely(__pyx_t_3 == -1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 39; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_Length(__pyx_t_2); if (unlikely(__pyx_t_3 == -1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 30; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __pyx_v_N = __pyx_t_3;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":40
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":31
  *         self.darray = DataArray(from_text=filename, side=side, use_sent_id=True)
  *         N = len(self.darray)
  *         V = len(self.darray.id2word)             # <<<<<<<<<<<<<<
@@ -33016,24 +33213,24 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
   __pyx_t_2 = __pyx_v_self->darray->id2word;
   __Pyx_INCREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_Length(__pyx_t_2); if (unlikely(__pyx_t_3 == -1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 40; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_Length(__pyx_t_2); if (unlikely(__pyx_t_3 == -1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 31; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __pyx_v_V = __pyx_t_3;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":42
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":33
  *         V = len(self.darray.id2word)
  * 
  *         self.sa = IntList(initial_len=N)             # <<<<<<<<<<<<<<
  *         self.ha = IntList(initial_len=V+1)
  * 
  */
-  __pyx_t_2 = PyDict_New(); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 42; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyDict_New(); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 33; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_2));
-  __pyx_t_1 = PyInt_FromLong(__pyx_v_N); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 42; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyInt_FromLong(__pyx_v_N); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 33; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  if (PyDict_SetItem(__pyx_t_2, ((PyObject *)__pyx_n_s__initial_len), __pyx_t_1) < 0) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 42; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_2, ((PyObject *)__pyx_n_s__initial_len), __pyx_t_1) < 0) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 33; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  __pyx_t_1 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_IntList)), ((PyObject *)__pyx_empty_tuple), ((PyObject *)__pyx_t_2)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 42; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_IntList)), ((PyObject *)__pyx_empty_tuple), ((PyObject *)__pyx_t_2)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 33; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
   __Pyx_GIVEREF(__pyx_t_1);
@@ -33042,20 +33239,20 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
   __pyx_v_self->sa = ((struct __pyx_obj_3_sa_IntList *)__pyx_t_1);
   __pyx_t_1 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":43
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":34
  * 
  *         self.sa = IntList(initial_len=N)
  *         self.ha = IntList(initial_len=V+1)             # <<<<<<<<<<<<<<
  * 
  *         isa = IntList(initial_len=N)
  */
-  __pyx_t_1 = PyDict_New(); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 43; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyDict_New(); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 34; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_1));
-  __pyx_t_2 = PyInt_FromLong((__pyx_v_V + 1)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 43; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyInt_FromLong((__pyx_v_V + 1)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 34; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_n_s__initial_len), __pyx_t_2) < 0) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 43; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_n_s__initial_len), __pyx_t_2) < 0) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 34; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_IntList)), ((PyObject *)__pyx_empty_tuple), ((PyObject *)__pyx_t_1)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 43; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_IntList)), ((PyObject *)__pyx_empty_tuple), ((PyObject *)__pyx_t_1)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 34; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
   __Pyx_GIVEREF(__pyx_t_2);
@@ -33064,45 +33261,45 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
   __pyx_v_self->ha = ((struct __pyx_obj_3_sa_IntList *)__pyx_t_2);
   __pyx_t_2 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":45
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":36
  *         self.ha = IntList(initial_len=V+1)
  * 
  *         isa = IntList(initial_len=N)             # <<<<<<<<<<<<<<
  *         word_count = IntList(initial_len=V+1)
  * 
  */
-  __pyx_t_2 = PyDict_New(); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 45; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyDict_New(); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 36; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_2));
-  __pyx_t_1 = PyInt_FromLong(__pyx_v_N); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 45; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyInt_FromLong(__pyx_v_N); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 36; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  if (PyDict_SetItem(__pyx_t_2, ((PyObject *)__pyx_n_s__initial_len), __pyx_t_1) < 0) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 45; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_2, ((PyObject *)__pyx_n_s__initial_len), __pyx_t_1) < 0) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 36; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  __pyx_t_1 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_IntList)), ((PyObject *)__pyx_empty_tuple), ((PyObject *)__pyx_t_2)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 45; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_IntList)), ((PyObject *)__pyx_empty_tuple), ((PyObject *)__pyx_t_2)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 36; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
   __pyx_v_isa = ((struct __pyx_obj_3_sa_IntList *)__pyx_t_1);
   __pyx_t_1 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":46
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":37
  * 
  *         isa = IntList(initial_len=N)
  *         word_count = IntList(initial_len=V+1)             # <<<<<<<<<<<<<<
  * 
  *         '''Step 1: bucket sort data'''
  */
-  __pyx_t_1 = PyDict_New(); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 46; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyDict_New(); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 37; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_1));
-  __pyx_t_2 = PyInt_FromLong((__pyx_v_V + 1)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 46; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyInt_FromLong((__pyx_v_V + 1)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 37; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_n_s__initial_len), __pyx_t_2) < 0) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 46; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_n_s__initial_len), __pyx_t_2) < 0) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 37; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_IntList)), ((PyObject *)__pyx_empty_tuple), ((PyObject *)__pyx_t_1)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 46; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_3_sa_IntList)), ((PyObject *)__pyx_empty_tuple), ((PyObject *)__pyx_t_1)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 37; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
   __pyx_v_word_count = ((struct __pyx_obj_3_sa_IntList *)__pyx_t_2);
   __pyx_t_2 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":49
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":40
  * 
  *         '''Step 1: bucket sort data'''
  *         cdef float sort_start_time = monitor_cpu()             # <<<<<<<<<<<<<<
@@ -33111,7 +33308,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
   __pyx_v_sort_start_time = __pyx_f_3_sa_monitor_cpu();
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":50
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":41
  *         '''Step 1: bucket sort data'''
  *         cdef float sort_start_time = monitor_cpu()
  *         cdef float start_time = sort_start_time             # <<<<<<<<<<<<<<
@@ -33120,7 +33317,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
   __pyx_v_start_time = __pyx_v_sort_start_time;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":51
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":42
  *         cdef float sort_start_time = monitor_cpu()
  *         cdef float start_time = sort_start_time
  *         for i from 0 <= i < N:             # <<<<<<<<<<<<<<
@@ -33130,7 +33327,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
   __pyx_t_4 = __pyx_v_N;
   for (__pyx_v_i = 0; __pyx_v_i < __pyx_t_4; __pyx_v_i++) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":52
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":43
  *         cdef float start_time = sort_start_time
  *         for i from 0 <= i < N:
  *             a_i = self.darray.data.arr[i]             # <<<<<<<<<<<<<<
@@ -33139,7 +33336,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
     __pyx_v_a_i = (__pyx_v_self->darray->data->arr[__pyx_v_i]);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":53
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":44
  *         for i from 0 <= i < N:
  *             a_i = self.darray.data.arr[i]
  *             word_count.arr[a_i] = word_count.arr[a_i] + 1             # <<<<<<<<<<<<<<
@@ -33149,7 +33346,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
     (__pyx_v_word_count->arr[__pyx_v_a_i]) = ((__pyx_v_word_count->arr[__pyx_v_a_i]) + 1);
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":55
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":46
  *             word_count.arr[a_i] = word_count.arr[a_i] + 1
  * 
  *         n = 0             # <<<<<<<<<<<<<<
@@ -33158,7 +33355,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
   __pyx_v_n = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":56
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":47
  * 
  *         n = 0
  *         for i from 0 <= i < V+1:             # <<<<<<<<<<<<<<
@@ -33168,7 +33365,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
   __pyx_t_5 = (__pyx_v_V + 1);
   for (__pyx_v_i = 0; __pyx_v_i < __pyx_t_5; __pyx_v_i++) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":57
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":48
  *         n = 0
  *         for i from 0 <= i < V+1:
  *             self.ha.arr[i] = n             # <<<<<<<<<<<<<<
@@ -33177,7 +33374,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
     (__pyx_v_self->ha->arr[__pyx_v_i]) = __pyx_v_n;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":58
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":49
  *         for i from 0 <= i < V+1:
  *             self.ha.arr[i] = n
  *             n = n + word_count.arr[i]             # <<<<<<<<<<<<<<
@@ -33186,7 +33383,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
     __pyx_v_n = (__pyx_v_n + (__pyx_v_word_count->arr[__pyx_v_i]));
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":59
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":50
  *             self.ha.arr[i] = n
  *             n = n + word_count.arr[i]
  *             word_count.arr[i] = 0             # <<<<<<<<<<<<<<
@@ -33196,7 +33393,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
     (__pyx_v_word_count->arr[__pyx_v_i]) = 0;
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":61
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":52
  *             word_count.arr[i] = 0
  * 
  *         for i from 0 <= i < N:             # <<<<<<<<<<<<<<
@@ -33206,7 +33403,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
   __pyx_t_4 = __pyx_v_N;
   for (__pyx_v_i = 0; __pyx_v_i < __pyx_t_4; __pyx_v_i++) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":62
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":53
  * 
  *         for i from 0 <= i < N:
  *             a_i = self.darray.data.arr[i]             # <<<<<<<<<<<<<<
@@ -33215,7 +33412,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
     __pyx_v_a_i = (__pyx_v_self->darray->data->arr[__pyx_v_i]);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":63
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":54
  *         for i from 0 <= i < N:
  *             a_i = self.darray.data.arr[i]
  *             self.sa.arr[self.ha.arr[a_i] + word_count.arr[a_i]] = i             # <<<<<<<<<<<<<<
@@ -33224,7 +33421,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
     (__pyx_v_self->sa->arr[((__pyx_v_self->ha->arr[__pyx_v_a_i]) + (__pyx_v_word_count->arr[__pyx_v_a_i]))]) = __pyx_v_i;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":64
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":55
  *             a_i = self.darray.data.arr[i]
  *             self.sa.arr[self.ha.arr[a_i] + word_count.arr[a_i]] = i
  *             isa.arr[i] = self.ha.arr[a_i + 1] - 1 # bucket pointer is last index in bucket             # <<<<<<<<<<<<<<
@@ -33233,7 +33430,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
     (__pyx_v_isa->arr[__pyx_v_i]) = ((__pyx_v_self->ha->arr[(__pyx_v_a_i + 1)]) - 1);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":65
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":56
  *             self.sa.arr[self.ha.arr[a_i] + word_count.arr[a_i]] = i
  *             isa.arr[i] = self.ha.arr[a_i + 1] - 1 # bucket pointer is last index in bucket
  *             word_count.arr[a_i] = word_count.arr[a_i] + 1             # <<<<<<<<<<<<<<
@@ -33243,7 +33440,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
     (__pyx_v_word_count->arr[__pyx_v_a_i]) = ((__pyx_v_word_count->arr[__pyx_v_a_i]) + 1);
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":68
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":59
  * 
  *         '''Determine size of initial runs'''
  *         current_run = 0             # <<<<<<<<<<<<<<
@@ -33252,7 +33449,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
   __pyx_v_current_run = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":69
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":60
  *         '''Determine size of initial runs'''
  *         current_run = 0
  *         for i from 0 <= i < V+1:             # <<<<<<<<<<<<<<
@@ -33262,7 +33459,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
   __pyx_t_5 = (__pyx_v_V + 1);
   for (__pyx_v_i = 0; __pyx_v_i < __pyx_t_5; __pyx_v_i++) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":70
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":61
  *         current_run = 0
  *         for i from 0 <= i < V+1:
  *             if i < V and self.ha.arr[i+1] - self.ha.arr[i] == 1:             # <<<<<<<<<<<<<<
@@ -33278,7 +33475,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
     }
     if (__pyx_t_8) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":71
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":62
  *         for i from 0 <= i < V+1:
  *             if i < V and self.ha.arr[i+1] - self.ha.arr[i] == 1:
  *                 current_run = current_run + 1             # <<<<<<<<<<<<<<
@@ -33290,7 +33487,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
     }
     /*else*/ {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":73
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":64
  *                 current_run = current_run + 1
  *             else:
  *                 if current_run > 0:             # <<<<<<<<<<<<<<
@@ -33300,7 +33497,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
       __pyx_t_8 = (__pyx_v_current_run > 0);
       if (__pyx_t_8) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":74
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":65
  *             else:
  *                 if current_run > 0:
  *                     self.sa.arr[self.ha.arr[i] - current_run] = -current_run             # <<<<<<<<<<<<<<
@@ -33309,7 +33506,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
         (__pyx_v_self->sa->arr[((__pyx_v_self->ha->arr[__pyx_v_i]) - __pyx_v_current_run)]) = (-__pyx_v_current_run);
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":75
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":66
  *                 if current_run > 0:
  *                     self.sa.arr[self.ha.arr[i] - current_run] = -current_run
  *                     current_run = 0             # <<<<<<<<<<<<<<
@@ -33324,35 +33521,35 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
     __pyx_L11:;
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":77
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":68
  *                     current_run = 0
  * 
  *         logger.info("    Bucket sort took %f seconds", (monitor_cpu() - sort_start_time))             # <<<<<<<<<<<<<<
  * 
  *         '''Step 2: prefix-doubling sort'''
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__logger); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 77; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__logger); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__info); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 77; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__info); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble((__pyx_f_3_sa_monitor_cpu() - __pyx_v_sort_start_time)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 77; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble((__pyx_f_3_sa_monitor_cpu() - __pyx_v_sort_start_time)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_9 = PyTuple_New(2); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 77; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_9 = PyTuple_New(2); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 68; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_9);
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_90));
-  PyTuple_SET_ITEM(__pyx_t_9, 0, ((PyObject *)__pyx_kp_s_90));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_90));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_89));
+  PyTuple_SET_ITEM(__pyx_t_9, 0, ((PyObject *)__pyx_kp_s_89));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_89));
   PyTuple_SET_ITEM(__pyx_t_9, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_9), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 77; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_9), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __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_9)); __pyx_t_9 = 0;
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":80
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":71
  * 
  *         '''Step 2: prefix-doubling sort'''
  *         h = 1             # <<<<<<<<<<<<<<
@@ -33361,7 +33558,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
   __pyx_v_h = 1;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":81
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":72
  *         '''Step 2: prefix-doubling sort'''
  *         h = 1
  *         while self.sa.arr[0] != -N:             # <<<<<<<<<<<<<<
@@ -33372,7 +33569,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
     __pyx_t_8 = ((__pyx_v_self->sa->arr[0]) != (-__pyx_v_N));
     if (!__pyx_t_8) break;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":82
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":73
  *         h = 1
  *         while self.sa.arr[0] != -N:
  *             sort_start_time = monitor_cpu()             # <<<<<<<<<<<<<<
@@ -33381,35 +33578,35 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
     __pyx_v_sort_start_time = __pyx_f_3_sa_monitor_cpu();
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":83
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":74
  *         while self.sa.arr[0] != -N:
  *             sort_start_time = monitor_cpu()
  *             logger.debug("    Refining, sort depth = %d", h)             # <<<<<<<<<<<<<<
  *             i = 0
  *             skip = 0
  */
-    __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__logger); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 83; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__logger); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 74; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_9 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__debug); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 83; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_9 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__debug); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 74; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_9);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    __pyx_t_2 = PyInt_FromLong(__pyx_v_h); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 83; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyInt_FromLong(__pyx_v_h); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 74; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_1 = PyTuple_New(2); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 83; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = PyTuple_New(2); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 74; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
-    __Pyx_INCREF(((PyObject *)__pyx_kp_s_91));
-    PyTuple_SET_ITEM(__pyx_t_1, 0, ((PyObject *)__pyx_kp_s_91));
-    __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_91));
+    __Pyx_INCREF(((PyObject *)__pyx_kp_s_90));
+    PyTuple_SET_ITEM(__pyx_t_1, 0, ((PyObject *)__pyx_kp_s_90));
+    __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_90));
     PyTuple_SET_ITEM(__pyx_t_1, 1, __pyx_t_2);
     __Pyx_GIVEREF(__pyx_t_2);
     __pyx_t_2 = 0;
-    __pyx_t_2 = PyObject_Call(__pyx_t_9, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 83; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_t_9, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 74; __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_1)); __pyx_t_1 = 0;
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":84
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":75
  *             sort_start_time = monitor_cpu()
  *             logger.debug("    Refining, sort depth = %d", h)
  *             i = 0             # <<<<<<<<<<<<<<
@@ -33418,7 +33615,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
     __pyx_v_i = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":85
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":76
  *             logger.debug("    Refining, sort depth = %d", h)
  *             i = 0
  *             skip = 0             # <<<<<<<<<<<<<<
@@ -33427,7 +33624,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
     __pyx_v_skip = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":86
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":77
  *             i = 0
  *             skip = 0
  *             while i < N:             # <<<<<<<<<<<<<<
@@ -33438,7 +33635,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
       __pyx_t_8 = (__pyx_v_i < __pyx_v_N);
       if (!__pyx_t_8) break;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":87
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":78
  *             skip = 0
  *             while i < N:
  *                 if self.sa.arr[i] < 0:             # <<<<<<<<<<<<<<
@@ -33448,7 +33645,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
       __pyx_t_8 = ((__pyx_v_self->sa->arr[__pyx_v_i]) < 0);
       if (__pyx_t_8) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":88
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":79
  *             while i < N:
  *                 if self.sa.arr[i] < 0:
  *                     skip = skip + self.sa.arr[i]             # <<<<<<<<<<<<<<
@@ -33457,7 +33654,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
         __pyx_v_skip = (__pyx_v_skip + (__pyx_v_self->sa->arr[__pyx_v_i]));
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":89
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":80
  *                 if self.sa.arr[i] < 0:
  *                     skip = skip + self.sa.arr[i]
  *                     i = i - self.sa.arr[i]             # <<<<<<<<<<<<<<
@@ -33469,7 +33666,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
       }
       /*else*/ {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":91
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":82
  *                     i = i - self.sa.arr[i]
  *                 else:
  *                     if skip < 0:             # <<<<<<<<<<<<<<
@@ -33479,7 +33676,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
         __pyx_t_8 = (__pyx_v_skip < 0);
         if (__pyx_t_8) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":92
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":83
  *                 else:
  *                     if skip < 0:
  *                         self.sa.arr[i+skip] = skip             # <<<<<<<<<<<<<<
@@ -33488,7 +33685,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
           (__pyx_v_self->sa->arr[(__pyx_v_i + __pyx_v_skip)]) = __pyx_v_skip;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":93
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":84
  *                     if skip < 0:
  *                         self.sa.arr[i+skip] = skip
  *                         skip = 0             # <<<<<<<<<<<<<<
@@ -33500,7 +33697,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
         }
         __pyx_L18:;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":94
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":85
  *                         self.sa.arr[i+skip] = skip
  *                         skip = 0
  *                     j = isa.arr[self.sa.arr[i]]             # <<<<<<<<<<<<<<
@@ -33509,22 +33706,22 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
         __pyx_v_j = (__pyx_v_isa->arr[(__pyx_v_self->sa->arr[__pyx_v_i])]);
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":95
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":86
  *                         skip = 0
  *                     j = isa.arr[self.sa.arr[i]]
  *                     self.q3sort(i, j, h, isa)             # <<<<<<<<<<<<<<
  *                     i = j+1
  *             if skip < 0:
  */
-        __pyx_t_2 = PyObject_GetAttr(((PyObject *)__pyx_v_self), __pyx_n_s__q3sort); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 95; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_2 = PyObject_GetAttr(((PyObject *)__pyx_v_self), __pyx_n_s__q3sort); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 86; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_2);
-        __pyx_t_1 = PyInt_FromLong(__pyx_v_i); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 95; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_1 = PyInt_FromLong(__pyx_v_i); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 86; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_1);
-        __pyx_t_9 = PyInt_FromLong(__pyx_v_j); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 95; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_9 = PyInt_FromLong(__pyx_v_j); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 86; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_9);
-        __pyx_t_10 = PyInt_FromLong(__pyx_v_h); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 95; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_10 = PyInt_FromLong(__pyx_v_h); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 86; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_10);
-        __pyx_t_11 = PyTuple_New(4); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 95; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_11 = PyTuple_New(4); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 86; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_11);
         PyTuple_SET_ITEM(__pyx_t_11, 0, __pyx_t_1);
         __Pyx_GIVEREF(__pyx_t_1);
@@ -33538,13 +33735,13 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
         __pyx_t_1 = 0;
         __pyx_t_9 = 0;
         __pyx_t_10 = 0;
-        __pyx_t_10 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_11), NULL); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 95; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __pyx_t_10 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_11), NULL); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 86; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_10);
         __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
         __Pyx_DECREF(((PyObject *)__pyx_t_11)); __pyx_t_11 = 0;
         __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":96
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":87
  *                     j = isa.arr[self.sa.arr[i]]
  *                     self.q3sort(i, j, h, isa)
  *                     i = j+1             # <<<<<<<<<<<<<<
@@ -33556,7 +33753,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
       __pyx_L17:;
     }
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":97
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":88
  *                     self.q3sort(i, j, h, isa)
  *                     i = j+1
  *             if skip < 0:             # <<<<<<<<<<<<<<
@@ -33566,7 +33763,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
     __pyx_t_8 = (__pyx_v_skip < 0);
     if (__pyx_t_8) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":98
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":89
  *                     i = j+1
  *             if skip < 0:
  *                 self.sa.arr[i+skip] = skip             # <<<<<<<<<<<<<<
@@ -33578,7 +33775,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
     }
     __pyx_L19:;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":99
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":90
  *             if skip < 0:
  *                 self.sa.arr[i+skip] = skip
  *             h = h * 2             # <<<<<<<<<<<<<<
@@ -33587,53 +33784,53 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
     __pyx_v_h = (__pyx_v_h * 2);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":100
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":91
  *                 self.sa.arr[i+skip] = skip
  *             h = h * 2
  *             logger.debug("    Refinement took %f seconds", (monitor_cpu() - sort_start_time))             # <<<<<<<<<<<<<<
  * 
  *         '''Step 3: read off suffix array from inverse suffix array'''
  */
-    __pyx_t_10 = __Pyx_GetName(__pyx_m, __pyx_n_s__logger); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 100; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_10 = __Pyx_GetName(__pyx_m, __pyx_n_s__logger); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 91; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_10);
-    __pyx_t_11 = PyObject_GetAttr(__pyx_t_10, __pyx_n_s__debug); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 100; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_11 = PyObject_GetAttr(__pyx_t_10, __pyx_n_s__debug); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 91; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_11);
     __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
-    __pyx_t_10 = PyFloat_FromDouble((__pyx_f_3_sa_monitor_cpu() - __pyx_v_sort_start_time)); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 100; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_10 = PyFloat_FromDouble((__pyx_f_3_sa_monitor_cpu() - __pyx_v_sort_start_time)); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 91; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_10);
-    __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 100; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 91; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_INCREF(((PyObject *)__pyx_kp_s_92));
-    PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_kp_s_92));
-    __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_92));
+    __Pyx_INCREF(((PyObject *)__pyx_kp_s_91));
+    PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_kp_s_91));
+    __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_91));
     PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_t_10);
     __Pyx_GIVEREF(__pyx_t_10);
     __pyx_t_10 = 0;
-    __pyx_t_10 = PyObject_Call(__pyx_t_11, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 100; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_10 = PyObject_Call(__pyx_t_11, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 91; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_10);
     __Pyx_DECREF(__pyx_t_11); __pyx_t_11 = 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/suffix_array.pxi":103
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":94
  * 
  *         '''Step 3: read off suffix array from inverse suffix array'''
  *         logger.info("    Finalizing sort...")             # <<<<<<<<<<<<<<
  *         for i from 0 <= i < N:
  *             j = isa.arr[i]
  */
-  __pyx_t_10 = __Pyx_GetName(__pyx_m, __pyx_n_s__logger); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 103; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_10 = __Pyx_GetName(__pyx_m, __pyx_n_s__logger); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 94; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_10);
-  __pyx_t_2 = PyObject_GetAttr(__pyx_t_10, __pyx_n_s__info); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 103; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_t_10, __pyx_n_s__info); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 94; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
-  __pyx_t_10 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_k_tuple_94), NULL); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 103; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_10 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_k_tuple_93), NULL); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 94; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_10);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":104
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":95
  *         '''Step 3: read off suffix array from inverse suffix array'''
  *         logger.info("    Finalizing sort...")
  *         for i from 0 <= i < N:             # <<<<<<<<<<<<<<
@@ -33643,7 +33840,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
   __pyx_t_4 = __pyx_v_N;
   for (__pyx_v_i = 0; __pyx_v_i < __pyx_t_4; __pyx_v_i++) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":105
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":96
  *         logger.info("    Finalizing sort...")
  *         for i from 0 <= i < N:
  *             j = isa.arr[i]             # <<<<<<<<<<<<<<
@@ -33652,7 +33849,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
  */
     __pyx_v_j = (__pyx_v_isa->arr[__pyx_v_i]);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":106
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":97
  *         for i from 0 <= i < N:
  *             j = isa.arr[i]
  *             self.sa.arr[j] = i             # <<<<<<<<<<<<<<
@@ -33662,29 +33859,29 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
     (__pyx_v_self->sa->arr[__pyx_v_j]) = __pyx_v_i;
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":107
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":98
  *             j = isa.arr[i]
  *             self.sa.arr[j] = i
  *         logger.info("Suffix array construction took %f seconds", (monitor_cpu() - start_time))             # <<<<<<<<<<<<<<
  * 
  *     def q3sort(self, int i, int j, int h, IntList isa, pad=""):
  */
-  __pyx_t_10 = __Pyx_GetName(__pyx_m, __pyx_n_s__logger); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_10 = __Pyx_GetName(__pyx_m, __pyx_n_s__logger); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 98; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_10);
-  __pyx_t_2 = PyObject_GetAttr(__pyx_t_10, __pyx_n_s__info); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_t_10, __pyx_n_s__info); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 98; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
-  __pyx_t_10 = PyFloat_FromDouble((__pyx_f_3_sa_monitor_cpu() - __pyx_v_start_time)); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_10 = PyFloat_FromDouble((__pyx_f_3_sa_monitor_cpu() - __pyx_v_start_time)); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 98; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_10);
-  __pyx_t_11 = PyTuple_New(2); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_11 = PyTuple_New(2); if (unlikely(!__pyx_t_11)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 98; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_11);
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_95));
-  PyTuple_SET_ITEM(__pyx_t_11, 0, ((PyObject *)__pyx_kp_s_95));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_95));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_94));
+  PyTuple_SET_ITEM(__pyx_t_11, 0, ((PyObject *)__pyx_kp_s_94));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_94));
   PyTuple_SET_ITEM(__pyx_t_11, 1, __pyx_t_10);
   __Pyx_GIVEREF(__pyx_t_10);
   __pyx_t_10 = 0;
-  __pyx_t_10 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_11), NULL); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_10 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_11), NULL); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 98; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_10);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_11)); __pyx_t_11 = 0;
@@ -33709,9 +33906,9 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_text(struct __pyx_obj_3_sa_S
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_11SuffixArray_13q3sort(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_3_sa_11SuffixArray_12q3sort[] = "This is a ternary quicksort. It divides the array into\n        three partitions: items less than the pivot, items equal\n        to pivot, and items greater than pivot.    The first and last\n        of these partitions are then recursively sorted";
-static PyObject *__pyx_pw_3_sa_11SuffixArray_13q3sort(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pw_3_sa_11SuffixArray_7q3sort(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_3_sa_11SuffixArray_6q3sort[] = "This is a ternary quicksort. It divides the array into\n        three partitions: items less than the pivot, items equal\n        to pivot, and items greater than pivot.    The first and last\n        of these partitions are then recursively sorted";
+static PyObject *__pyx_pw_3_sa_11SuffixArray_7q3sort(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   int __pyx_v_i;
   int __pyx_v_j;
   int __pyx_v_h;
@@ -33744,17 +33941,17 @@ static PyObject *__pyx_pw_3_sa_11SuffixArray_13q3sort(PyObject *__pyx_v_self, Py
         case  1:
         if (likely((values[1] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__j)) != 0)) kw_args--;
         else {
-          __Pyx_RaiseArgtupleInvalid("q3sort", 0, 4, 5, 1); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 109; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+          __Pyx_RaiseArgtupleInvalid("q3sort", 0, 4, 5, 1); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 100; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
         }
         case  2:
         if (likely((values[2] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__h)) != 0)) kw_args--;
         else {
-          __Pyx_RaiseArgtupleInvalid("q3sort", 0, 4, 5, 2); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 109; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+          __Pyx_RaiseArgtupleInvalid("q3sort", 0, 4, 5, 2); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 100; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
         }
         case  3:
         if (likely((values[3] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__isa)) != 0)) kw_args--;
         else {
-          __Pyx_RaiseArgtupleInvalid("q3sort", 0, 4, 5, 3); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 109; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+          __Pyx_RaiseArgtupleInvalid("q3sort", 0, 4, 5, 3); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 100; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
         }
         case  4:
         if (kw_args > 0) {
@@ -33763,7 +33960,7 @@ static PyObject *__pyx_pw_3_sa_11SuffixArray_13q3sort(PyObject *__pyx_v_self, Py
         }
       }
       if (unlikely(kw_args > 0)) {
-        if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "q3sort") < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 109; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+        if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "q3sort") < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 100; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
     } else {
       switch (PyTuple_GET_SIZE(__pyx_args)) {
@@ -33776,22 +33973,22 @@ static PyObject *__pyx_pw_3_sa_11SuffixArray_13q3sort(PyObject *__pyx_v_self, Py
         default: goto __pyx_L5_argtuple_error;
       }
     }
-    __pyx_v_i = __Pyx_PyInt_AsInt(values[0]); if (unlikely((__pyx_v_i == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 109; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
-    __pyx_v_j = __Pyx_PyInt_AsInt(values[1]); if (unlikely((__pyx_v_j == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 109; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
-    __pyx_v_h = __Pyx_PyInt_AsInt(values[2]); if (unlikely((__pyx_v_h == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 109; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_v_i = __Pyx_PyInt_AsInt(values[0]); if (unlikely((__pyx_v_i == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 100; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_v_j = __Pyx_PyInt_AsInt(values[1]); if (unlikely((__pyx_v_j == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 100; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_v_h = __Pyx_PyInt_AsInt(values[2]); if (unlikely((__pyx_v_h == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 100; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     __pyx_v_isa = ((struct __pyx_obj_3_sa_IntList *)values[3]);
     __pyx_v_pad = values[4];
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("q3sort", 0, 4, 5, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 109; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("q3sort", 0, 4, 5, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 100; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
   __Pyx_AddTraceback("_sa.SuffixArray.q3sort", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  if (unlikely(!__Pyx_ArgTypeTest(((PyObject *)__pyx_v_isa), __pyx_ptype_3_sa_IntList, 1, "isa", 0))) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 109; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __pyx_r = __pyx_pf_3_sa_11SuffixArray_12q3sort(((struct __pyx_obj_3_sa_SuffixArray *)__pyx_v_self), __pyx_v_i, __pyx_v_j, __pyx_v_h, __pyx_v_isa, __pyx_v_pad);
+  if (unlikely(!__Pyx_ArgTypeTest(((PyObject *)__pyx_v_isa), __pyx_ptype_3_sa_IntList, 1, "isa", 0))) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 100; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_r = __pyx_pf_3_sa_11SuffixArray_6q3sort(((struct __pyx_obj_3_sa_SuffixArray *)__pyx_v_self), __pyx_v_i, __pyx_v_j, __pyx_v_h, __pyx_v_isa, __pyx_v_pad);
   goto __pyx_L0;
   __pyx_L1_error:;
   __pyx_r = NULL;
@@ -33800,7 +33997,7 @@ static PyObject *__pyx_pw_3_sa_11SuffixArray_13q3sort(PyObject *__pyx_v_self, Py
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":109
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":100
  *         logger.info("Suffix array construction took %f seconds", (monitor_cpu() - start_time))
  * 
  *     def q3sort(self, int i, int j, int h, IntList isa, pad=""):             # <<<<<<<<<<<<<<
@@ -33808,7 +34005,7 @@ static PyObject *__pyx_pw_3_sa_11SuffixArray_13q3sort(PyObject *__pyx_v_self, Py
  *         three partitions: items less than the pivot, items equal
  */
 
-static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, int __pyx_v_i, int __pyx_v_j, int __pyx_v_h, struct __pyx_obj_3_sa_IntList *__pyx_v_isa, PyObject *__pyx_v_pad) {
+static PyObject *__pyx_pf_3_sa_11SuffixArray_6q3sort(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, int __pyx_v_i, int __pyx_v_j, int __pyx_v_h, struct __pyx_obj_3_sa_IntList *__pyx_v_isa, PyObject *__pyx_v_pad) {
   int __pyx_v_k;
   int __pyx_v_midpoint;
   int __pyx_v_pval;
@@ -33830,7 +34027,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("q3sort", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":116
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":107
  *         cdef int k, midpoint, pval, phead, ptail, tmp
  * 
  *         if j-i < -1:             # <<<<<<<<<<<<<<
@@ -33840,18 +34037,18 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
   __pyx_t_1 = ((__pyx_v_j - __pyx_v_i) < -1);
   if (__pyx_t_1) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":117
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":108
  * 
  *         if j-i < -1:
  *             raise Exception("Unexpected condition found in q3sort: sort from %d to %d" % (i,j))             # <<<<<<<<<<<<<<
  *         if j-i == -1:    # recursive base case -- empty interval
  *             return
  */
-    __pyx_t_2 = PyInt_FromLong(__pyx_v_i); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 117; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyInt_FromLong(__pyx_v_i); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 108; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_3 = PyInt_FromLong(__pyx_v_j); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 117; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyInt_FromLong(__pyx_v_j); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 108; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
-    __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 117; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 108; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
     PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_2);
     __Pyx_GIVEREF(__pyx_t_2);
@@ -33859,25 +34056,25 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
     __Pyx_GIVEREF(__pyx_t_3);
     __pyx_t_2 = 0;
     __pyx_t_3 = 0;
-    __pyx_t_3 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_96), ((PyObject *)__pyx_t_4)); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 117; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_95), ((PyObject *)__pyx_t_4)); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 108; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_3));
     __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-    __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 117; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 108; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
     PyTuple_SET_ITEM(__pyx_t_4, 0, ((PyObject *)__pyx_t_3));
     __Pyx_GIVEREF(((PyObject *)__pyx_t_3));
     __pyx_t_3 = 0;
-    __pyx_t_3 = PyObject_Call(__pyx_builtin_Exception, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 117; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyObject_Call(__pyx_builtin_Exception, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 108; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
     __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
     __Pyx_Raise(__pyx_t_3, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-    {__pyx_filename = __pyx_f[12]; __pyx_lineno = 117; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[12]; __pyx_lineno = 108; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L3;
   }
   __pyx_L3:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":118
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":109
  *         if j-i < -1:
  *             raise Exception("Unexpected condition found in q3sort: sort from %d to %d" % (i,j))
  *         if j-i == -1:    # recursive base case -- empty interval             # <<<<<<<<<<<<<<
@@ -33887,7 +34084,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
   __pyx_t_1 = ((__pyx_v_j - __pyx_v_i) == -1);
   if (__pyx_t_1) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":119
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":110
  *             raise Exception("Unexpected condition found in q3sort: sort from %d to %d" % (i,j))
  *         if j-i == -1:    # recursive base case -- empty interval
  *             return             # <<<<<<<<<<<<<<
@@ -33901,7 +34098,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
   }
   __pyx_L4:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":120
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":111
  *         if j-i == -1:    # recursive base case -- empty interval
  *             return
  *         if (j-i == 0):    # recursive base case -- singleton interval             # <<<<<<<<<<<<<<
@@ -33911,7 +34108,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
   __pyx_t_1 = ((__pyx_v_j - __pyx_v_i) == 0);
   if (__pyx_t_1) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":121
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":112
  *             return
  *         if (j-i == 0):    # recursive base case -- singleton interval
  *             isa.arr[self.sa.arr[i]] = i             # <<<<<<<<<<<<<<
@@ -33920,7 +34117,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
  */
     (__pyx_v_isa->arr[(__pyx_v_self->sa->arr[__pyx_v_i])]) = __pyx_v_i;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":122
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":113
  *         if (j-i == 0):    # recursive base case -- singleton interval
  *             isa.arr[self.sa.arr[i]] = i
  *             self.sa.arr[i] = -1             # <<<<<<<<<<<<<<
@@ -33929,7 +34126,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
  */
     (__pyx_v_self->sa->arr[__pyx_v_i]) = -1;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":123
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":114
  *             isa.arr[self.sa.arr[i]] = i
  *             self.sa.arr[i] = -1
  *             return             # <<<<<<<<<<<<<<
@@ -33943,7 +34140,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
   }
   __pyx_L5:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":132
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":123
  *         # If the method of assigning word_id's is changed, this method
  *         # may need to be reconsidered as well.
  *         midpoint = (i+j)/2             # <<<<<<<<<<<<<<
@@ -33952,7 +34149,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
  */
   __pyx_v_midpoint = __Pyx_div_long((__pyx_v_i + __pyx_v_j), 2);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":133
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":124
  *         # may need to be reconsidered as well.
  *         midpoint = (i+j)/2
  *         pval = isa.arr[self.sa.arr[midpoint] + h]             # <<<<<<<<<<<<<<
@@ -33961,7 +34158,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
  */
   __pyx_v_pval = (__pyx_v_isa->arr[((__pyx_v_self->sa->arr[__pyx_v_midpoint]) + __pyx_v_h)]);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":134
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":125
  *         midpoint = (i+j)/2
  *         pval = isa.arr[self.sa.arr[midpoint] + h]
  *         if i != midpoint:             # <<<<<<<<<<<<<<
@@ -33971,7 +34168,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
   __pyx_t_1 = (__pyx_v_i != __pyx_v_midpoint);
   if (__pyx_t_1) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":135
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":126
  *         pval = isa.arr[self.sa.arr[midpoint] + h]
  *         if i != midpoint:
  *             tmp = self.sa.arr[midpoint]             # <<<<<<<<<<<<<<
@@ -33980,7 +34177,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
  */
     __pyx_v_tmp = (__pyx_v_self->sa->arr[__pyx_v_midpoint]);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":136
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":127
  *         if i != midpoint:
  *             tmp = self.sa.arr[midpoint]
  *             self.sa.arr[midpoint] = self.sa.arr[i]             # <<<<<<<<<<<<<<
@@ -33989,7 +34186,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
  */
     (__pyx_v_self->sa->arr[__pyx_v_midpoint]) = (__pyx_v_self->sa->arr[__pyx_v_i]);
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":137
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":128
  *             tmp = self.sa.arr[midpoint]
  *             self.sa.arr[midpoint] = self.sa.arr[i]
  *             self.sa.arr[i] = tmp             # <<<<<<<<<<<<<<
@@ -34001,7 +34198,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
   }
   __pyx_L6:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":138
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":129
  *             self.sa.arr[midpoint] = self.sa.arr[i]
  *             self.sa.arr[i] = tmp
  *         phead = i             # <<<<<<<<<<<<<<
@@ -34010,7 +34207,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
  */
   __pyx_v_phead = __pyx_v_i;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":139
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":130
  *             self.sa.arr[i] = tmp
  *         phead = i
  *         ptail = i             # <<<<<<<<<<<<<<
@@ -34019,7 +34216,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
  */
   __pyx_v_ptail = __pyx_v_i;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":143
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":134
  *         # find the three partitions.    phead marks the first element
  *         # of the middle partition, and ptail marks the last element
  *         for k from i+1 <= k < j+1:             # <<<<<<<<<<<<<<
@@ -34029,7 +34226,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
   __pyx_t_5 = (__pyx_v_j + 1);
   for (__pyx_v_k = (__pyx_v_i + 1); __pyx_v_k < __pyx_t_5; __pyx_v_k++) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":144
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":135
  *         # of the middle partition, and ptail marks the last element
  *         for k from i+1 <= k < j+1:
  *             if isa.arr[self.sa.arr[k] + h] < pval:             # <<<<<<<<<<<<<<
@@ -34039,7 +34236,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
     __pyx_t_1 = ((__pyx_v_isa->arr[((__pyx_v_self->sa->arr[__pyx_v_k]) + __pyx_v_h)]) < __pyx_v_pval);
     if (__pyx_t_1) {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":145
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":136
  *         for k from i+1 <= k < j+1:
  *             if isa.arr[self.sa.arr[k] + h] < pval:
  *                 if k > ptail+1:             # <<<<<<<<<<<<<<
@@ -34049,7 +34246,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
       __pyx_t_1 = (__pyx_v_k > (__pyx_v_ptail + 1));
       if (__pyx_t_1) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":146
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":137
  *             if isa.arr[self.sa.arr[k] + h] < pval:
  *                 if k > ptail+1:
  *                     tmp = self.sa.arr[phead]             # <<<<<<<<<<<<<<
@@ -34058,7 +34255,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
  */
         __pyx_v_tmp = (__pyx_v_self->sa->arr[__pyx_v_phead]);
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":147
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":138
  *                 if k > ptail+1:
  *                     tmp = self.sa.arr[phead]
  *                     self.sa.arr[phead] = self.sa.arr[k]             # <<<<<<<<<<<<<<
@@ -34067,7 +34264,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
  */
         (__pyx_v_self->sa->arr[__pyx_v_phead]) = (__pyx_v_self->sa->arr[__pyx_v_k]);
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":148
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":139
  *                     tmp = self.sa.arr[phead]
  *                     self.sa.arr[phead] = self.sa.arr[k]
  *                     self.sa.arr[k] = self.sa.arr[ptail+1]             # <<<<<<<<<<<<<<
@@ -34076,7 +34273,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
  */
         (__pyx_v_self->sa->arr[__pyx_v_k]) = (__pyx_v_self->sa->arr[(__pyx_v_ptail + 1)]);
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":149
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":140
  *                     self.sa.arr[phead] = self.sa.arr[k]
  *                     self.sa.arr[k] = self.sa.arr[ptail+1]
  *                     self.sa.arr[ptail+1] = tmp             # <<<<<<<<<<<<<<
@@ -34088,7 +34285,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
       }
       /*else*/ {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":151
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":142
  *                     self.sa.arr[ptail+1] = tmp
  *                 else: # k == ptail+1
  *                     tmp = self.sa.arr[phead]             # <<<<<<<<<<<<<<
@@ -34097,7 +34294,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
  */
         __pyx_v_tmp = (__pyx_v_self->sa->arr[__pyx_v_phead]);
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":152
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":143
  *                 else: # k == ptail+1
  *                     tmp = self.sa.arr[phead]
  *                     self.sa.arr[phead] = self.sa.arr[k]             # <<<<<<<<<<<<<<
@@ -34106,7 +34303,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
  */
         (__pyx_v_self->sa->arr[__pyx_v_phead]) = (__pyx_v_self->sa->arr[__pyx_v_k]);
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":153
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":144
  *                     tmp = self.sa.arr[phead]
  *                     self.sa.arr[phead] = self.sa.arr[k]
  *                     self.sa.arr[k] = tmp             # <<<<<<<<<<<<<<
@@ -34117,7 +34314,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
       }
       __pyx_L10:;
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":154
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":145
  *                     self.sa.arr[phead] = self.sa.arr[k]
  *                     self.sa.arr[k] = tmp
  *                 phead = phead + 1             # <<<<<<<<<<<<<<
@@ -34126,7 +34323,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
  */
       __pyx_v_phead = (__pyx_v_phead + 1);
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":155
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":146
  *                     self.sa.arr[k] = tmp
  *                 phead = phead + 1
  *                 ptail = ptail + 1             # <<<<<<<<<<<<<<
@@ -34138,7 +34335,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
     }
     /*else*/ {
 
-      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":157
+      /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":148
  *                 ptail = ptail + 1
  *             else:
  *                 if isa.arr[self.sa.arr[k] + h] == pval:             # <<<<<<<<<<<<<<
@@ -34148,7 +34345,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
       __pyx_t_1 = ((__pyx_v_isa->arr[((__pyx_v_self->sa->arr[__pyx_v_k]) + __pyx_v_h)]) == __pyx_v_pval);
       if (__pyx_t_1) {
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":158
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":149
  *             else:
  *                 if isa.arr[self.sa.arr[k] + h] == pval:
  *                     if k > ptail+1:             # <<<<<<<<<<<<<<
@@ -34158,7 +34355,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
         __pyx_t_1 = (__pyx_v_k > (__pyx_v_ptail + 1));
         if (__pyx_t_1) {
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":159
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":150
  *                 if isa.arr[self.sa.arr[k] + h] == pval:
  *                     if k > ptail+1:
  *                         tmp = self.sa.arr[ptail+1]             # <<<<<<<<<<<<<<
@@ -34167,7 +34364,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
  */
           __pyx_v_tmp = (__pyx_v_self->sa->arr[(__pyx_v_ptail + 1)]);
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":160
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":151
  *                     if k > ptail+1:
  *                         tmp = self.sa.arr[ptail+1]
  *                         self.sa.arr[ptail+1] = self.sa.arr[k]             # <<<<<<<<<<<<<<
@@ -34176,7 +34373,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
  */
           (__pyx_v_self->sa->arr[(__pyx_v_ptail + 1)]) = (__pyx_v_self->sa->arr[__pyx_v_k]);
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":161
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":152
  *                         tmp = self.sa.arr[ptail+1]
  *                         self.sa.arr[ptail+1] = self.sa.arr[k]
  *                         self.sa.arr[k] = tmp             # <<<<<<<<<<<<<<
@@ -34188,7 +34385,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
         }
         __pyx_L12:;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":162
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":153
  *                         self.sa.arr[ptail+1] = self.sa.arr[k]
  *                         self.sa.arr[k] = tmp
  *                     ptail = ptail + 1             # <<<<<<<<<<<<<<
@@ -34203,24 +34400,24 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
     __pyx_L9:;
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":165
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":156
  * 
  *         # recursively sort smaller suffixes
  *         self.q3sort(i, phead-1, h, isa, pad+"    ")             # <<<<<<<<<<<<<<
  * 
  *         # update suffixes with pivot value
  */
-  __pyx_t_3 = PyObject_GetAttr(((PyObject *)__pyx_v_self), __pyx_n_s__q3sort); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 165; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(((PyObject *)__pyx_v_self), __pyx_n_s__q3sort); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 156; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
-  __pyx_t_4 = PyInt_FromLong(__pyx_v_i); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 165; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyInt_FromLong(__pyx_v_i); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 156; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_2 = PyInt_FromLong((__pyx_v_phead - 1)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 165; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyInt_FromLong((__pyx_v_phead - 1)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 156; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_6 = PyInt_FromLong(__pyx_v_h); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 165; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_6 = PyInt_FromLong(__pyx_v_h); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 156; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_6);
-  __pyx_t_7 = PyNumber_Add(__pyx_v_pad, ((PyObject *)__pyx_kp_s_48)); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 165; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_7 = PyNumber_Add(__pyx_v_pad, ((PyObject *)__pyx_kp_s_48)); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 156; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_7);
-  __pyx_t_8 = PyTuple_New(5); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 165; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_8 = PyTuple_New(5); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 156; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_8);
   PyTuple_SET_ITEM(__pyx_t_8, 0, __pyx_t_4);
   __Pyx_GIVEREF(__pyx_t_4);
@@ -34237,13 +34434,13 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
   __pyx_t_2 = 0;
   __pyx_t_6 = 0;
   __pyx_t_7 = 0;
-  __pyx_t_7 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_8), NULL); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 165; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_7 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_8), NULL); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 156; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_7);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_8)); __pyx_t_8 = 0;
   __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":169
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":160
  *         # update suffixes with pivot value
  *         # corresponds to update_group function in Larsson & Sadakane
  *         for k from phead <= k < ptail+1:             # <<<<<<<<<<<<<<
@@ -34253,7 +34450,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
   __pyx_t_5 = (__pyx_v_ptail + 1);
   for (__pyx_v_k = __pyx_v_phead; __pyx_v_k < __pyx_t_5; __pyx_v_k++) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":170
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":161
  *         # corresponds to update_group function in Larsson & Sadakane
  *         for k from phead <= k < ptail+1:
  *             isa.arr[self.sa.arr[k]] = ptail             # <<<<<<<<<<<<<<
@@ -34263,7 +34460,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
     (__pyx_v_isa->arr[(__pyx_v_self->sa->arr[__pyx_v_k])]) = __pyx_v_ptail;
   }
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":171
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":162
  *         for k from phead <= k < ptail+1:
  *             isa.arr[self.sa.arr[k]] = ptail
  *         if phead == ptail:             # <<<<<<<<<<<<<<
@@ -34273,7 +34470,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
   __pyx_t_1 = (__pyx_v_phead == __pyx_v_ptail);
   if (__pyx_t_1) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":172
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":163
  *             isa.arr[self.sa.arr[k]] = ptail
  *         if phead == ptail:
  *             self.sa.arr[phead] = -1             # <<<<<<<<<<<<<<
@@ -34285,24 +34482,24 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
   }
   __pyx_L15:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":175
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":166
  * 
  *         # recursively sort larger suffixes
  *         self.q3sort(ptail+1, j, h, isa, pad+"    ")             # <<<<<<<<<<<<<<
  * 
  * 
  */
-  __pyx_t_7 = PyObject_GetAttr(((PyObject *)__pyx_v_self), __pyx_n_s__q3sort); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 175; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_7 = PyObject_GetAttr(((PyObject *)__pyx_v_self), __pyx_n_s__q3sort); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 166; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_7);
-  __pyx_t_8 = PyInt_FromLong((__pyx_v_ptail + 1)); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 175; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_8 = PyInt_FromLong((__pyx_v_ptail + 1)); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 166; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_8);
-  __pyx_t_3 = PyInt_FromLong(__pyx_v_j); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 175; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyInt_FromLong(__pyx_v_j); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 166; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
-  __pyx_t_6 = PyInt_FromLong(__pyx_v_h); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 175; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_6 = PyInt_FromLong(__pyx_v_h); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 166; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_6);
-  __pyx_t_2 = PyNumber_Add(__pyx_v_pad, ((PyObject *)__pyx_kp_s_48)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 175; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyNumber_Add(__pyx_v_pad, ((PyObject *)__pyx_kp_s_48)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 166; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyTuple_New(5); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 175; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(5); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 166; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_8);
   __Pyx_GIVEREF(__pyx_t_8);
@@ -34319,7 +34516,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
   __pyx_t_3 = 0;
   __pyx_t_6 = 0;
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_7, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 175; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_7, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 166; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
@@ -34343,14 +34540,14 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_12q3sort(struct __pyx_obj_3_sa_Suff
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_11SuffixArray_15write_text(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename); /*proto*/
-static PyObject *__pyx_pw_3_sa_11SuffixArray_15write_text(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename) {
+static PyObject *__pyx_pw_3_sa_11SuffixArray_9write_text(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename); /*proto*/
+static PyObject *__pyx_pw_3_sa_11SuffixArray_9write_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[12]; __pyx_lineno = 178; __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[12]; __pyx_lineno = 169; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L3_error:;
@@ -34358,12 +34555,12 @@ static PyObject *__pyx_pw_3_sa_11SuffixArray_15write_text(PyObject *__pyx_v_self
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_r = __pyx_pf_3_sa_11SuffixArray_14write_text(((struct __pyx_obj_3_sa_SuffixArray *)__pyx_v_self), ((char *)__pyx_v_filename));
+  __pyx_r = __pyx_pf_3_sa_11SuffixArray_8write_text(((struct __pyx_obj_3_sa_SuffixArray *)__pyx_v_self), ((char *)__pyx_v_filename));
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":178
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":169
  * 
  * 
  *     def write_text(self, char* filename):             # <<<<<<<<<<<<<<
@@ -34371,7 +34568,7 @@ static PyObject *__pyx_pw_3_sa_11SuffixArray_15write_text(PyObject *__pyx_v_self
  * 
  */
 
-static PyObject *__pyx_pf_3_sa_11SuffixArray_14write_text(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, char *__pyx_v_filename) {
+static PyObject *__pyx_pf_3_sa_11SuffixArray_8write_text(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, char *__pyx_v_filename) {
   PyObject *__pyx_r = NULL;
   __Pyx_RefNannyDeclarations
   PyObject *__pyx_t_1 = NULL;
@@ -34382,23 +34579,23 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_14write_text(struct __pyx_obj_3_sa_
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("write_text", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":179
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":170
  * 
  *     def write_text(self, char* filename):
  *         self.darray.write_text(filename)             # <<<<<<<<<<<<<<
  * 
  *     def read_binary(self, char* filename):
  */
-  __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self->darray), __pyx_n_s__write_text); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 179; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self->darray), __pyx_n_s__write_text); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 170; __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[12]; __pyx_lineno = 179; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyBytes_FromString(__pyx_v_filename); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 170; __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[12]; __pyx_lineno = 179; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 170; __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[12]; __pyx_lineno = 179; __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[12]; __pyx_lineno = 170; __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;
@@ -34419,14 +34616,14 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_14write_text(struct __pyx_obj_3_sa_
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_11SuffixArray_17read_binary(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename); /*proto*/
-static PyObject *__pyx_pw_3_sa_11SuffixArray_17read_binary(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename) {
+static PyObject *__pyx_pw_3_sa_11SuffixArray_11read_binary(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename); /*proto*/
+static PyObject *__pyx_pw_3_sa_11SuffixArray_11read_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[12]; __pyx_lineno = 181; __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[12]; __pyx_lineno = 172; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L3_error:;
@@ -34434,12 +34631,12 @@ static PyObject *__pyx_pw_3_sa_11SuffixArray_17read_binary(PyObject *__pyx_v_sel
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_r = __pyx_pf_3_sa_11SuffixArray_16read_binary(((struct __pyx_obj_3_sa_SuffixArray *)__pyx_v_self), ((char *)__pyx_v_filename));
+  __pyx_r = __pyx_pf_3_sa_11SuffixArray_10read_binary(((struct __pyx_obj_3_sa_SuffixArray *)__pyx_v_self), ((char *)__pyx_v_filename));
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":181
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":172
  *         self.darray.write_text(filename)
  * 
  *     def read_binary(self, char* filename):             # <<<<<<<<<<<<<<
@@ -34447,13 +34644,13 @@ static PyObject *__pyx_pw_3_sa_11SuffixArray_17read_binary(PyObject *__pyx_v_sel
  *         f = fopen(filename, "r")
  */
 
-static PyObject *__pyx_pf_3_sa_11SuffixArray_16read_binary(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, char *__pyx_v_filename) {
+static PyObject *__pyx_pf_3_sa_11SuffixArray_10read_binary(struct __pyx_obj_3_sa_SuffixArray *__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/suffix_array.pxi":183
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":174
  *     def read_binary(self, char* filename):
  *         cdef FILE *f
  *         f = fopen(filename, "r")             # <<<<<<<<<<<<<<
@@ -34462,7 +34659,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_16read_binary(struct __pyx_obj_3_sa
  */
   __pyx_v_f = fopen(__pyx_v_filename, __pyx_k__r);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":184
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":175
  *         cdef FILE *f
  *         f = fopen(filename, "r")
  *         self.darray.read_handle(f)             # <<<<<<<<<<<<<<
@@ -34471,7 +34668,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_16read_binary(struct __pyx_obj_3_sa
  */
   ((struct __pyx_vtabstruct_3_sa_DataArray *)__pyx_v_self->darray->__pyx_vtab)->read_handle(__pyx_v_self->darray, __pyx_v_f);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":185
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":176
  *         f = fopen(filename, "r")
  *         self.darray.read_handle(f)
  *         self.sa.read_handle(f)             # <<<<<<<<<<<<<<
@@ -34480,7 +34677,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_16read_binary(struct __pyx_obj_3_sa
  */
   ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_self->sa->__pyx_vtab)->read_handle(__pyx_v_self->sa, __pyx_v_f);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":186
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":177
  *         self.darray.read_handle(f)
  *         self.sa.read_handle(f)
  *         self.ha.read_handle(f)             # <<<<<<<<<<<<<<
@@ -34489,7 +34686,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_16read_binary(struct __pyx_obj_3_sa
  */
   ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_self->ha->__pyx_vtab)->read_handle(__pyx_v_self->ha, __pyx_v_f);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":187
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":178
  *         self.sa.read_handle(f)
  *         self.ha.read_handle(f)
  *         fclose(f)             # <<<<<<<<<<<<<<
@@ -34505,14 +34702,14 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_16read_binary(struct __pyx_obj_3_sa
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_11SuffixArray_19write_binary(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename); /*proto*/
-static PyObject *__pyx_pw_3_sa_11SuffixArray_19write_binary(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename) {
+static PyObject *__pyx_pw_3_sa_11SuffixArray_13write_binary(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename); /*proto*/
+static PyObject *__pyx_pw_3_sa_11SuffixArray_13write_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[12]; __pyx_lineno = 189; __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[12]; __pyx_lineno = 180; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L3_error:;
@@ -34520,12 +34717,12 @@ static PyObject *__pyx_pw_3_sa_11SuffixArray_19write_binary(PyObject *__pyx_v_se
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_r = __pyx_pf_3_sa_11SuffixArray_18write_binary(((struct __pyx_obj_3_sa_SuffixArray *)__pyx_v_self), ((char *)__pyx_v_filename));
+  __pyx_r = __pyx_pf_3_sa_11SuffixArray_12write_binary(((struct __pyx_obj_3_sa_SuffixArray *)__pyx_v_self), ((char *)__pyx_v_filename));
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":189
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":180
  *         fclose(f)
  * 
  *     def write_binary(self, char* filename):             # <<<<<<<<<<<<<<
@@ -34533,13 +34730,13 @@ static PyObject *__pyx_pw_3_sa_11SuffixArray_19write_binary(PyObject *__pyx_v_se
  *         f = fopen(filename, "w")
  */
 
-static PyObject *__pyx_pf_3_sa_11SuffixArray_18write_binary(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, char *__pyx_v_filename) {
+static PyObject *__pyx_pf_3_sa_11SuffixArray_12write_binary(struct __pyx_obj_3_sa_SuffixArray *__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/suffix_array.pxi":191
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":182
  *     def write_binary(self, char* filename):
  *         cdef FILE* f
  *         f = fopen(filename, "w")             # <<<<<<<<<<<<<<
@@ -34548,7 +34745,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_18write_binary(struct __pyx_obj_3_s
  */
   __pyx_v_f = fopen(__pyx_v_filename, __pyx_k__w);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":192
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":183
  *         cdef FILE* f
  *         f = fopen(filename, "w")
  *         self.darray.write_handle(f)             # <<<<<<<<<<<<<<
@@ -34557,7 +34754,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_18write_binary(struct __pyx_obj_3_s
  */
   ((struct __pyx_vtabstruct_3_sa_DataArray *)__pyx_v_self->darray->__pyx_vtab)->write_handle(__pyx_v_self->darray, __pyx_v_f);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":193
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":184
  *         f = fopen(filename, "w")
  *         self.darray.write_handle(f)
  *         self.sa.write_handle(f)             # <<<<<<<<<<<<<<
@@ -34566,7 +34763,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_18write_binary(struct __pyx_obj_3_s
  */
   ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_self->sa->__pyx_vtab)->write_handle(__pyx_v_self->sa, __pyx_v_f);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":194
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":185
  *         self.darray.write_handle(f)
  *         self.sa.write_handle(f)
  *         self.ha.write_handle(f)             # <<<<<<<<<<<<<<
@@ -34575,7 +34772,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_18write_binary(struct __pyx_obj_3_s
  */
   ((struct __pyx_vtabstruct_3_sa_IntList *)__pyx_v_self->ha->__pyx_vtab)->write_handle(__pyx_v_self->ha, __pyx_v_f);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":195
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":186
  *         self.sa.write_handle(f)
  *         self.ha.write_handle(f)
  *         fclose(f)             # <<<<<<<<<<<<<<
@@ -34591,14 +34788,14 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_18write_binary(struct __pyx_obj_3_s
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_11SuffixArray_21write_enhanced(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename); /*proto*/
-static PyObject *__pyx_pw_3_sa_11SuffixArray_21write_enhanced(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename) {
+static PyObject *__pyx_pw_3_sa_11SuffixArray_15write_enhanced(PyObject *__pyx_v_self, PyObject *__pyx_arg_filename); /*proto*/
+static PyObject *__pyx_pw_3_sa_11SuffixArray_15write_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[12]; __pyx_lineno = 197; __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[12]; __pyx_lineno = 188; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L3_error:;
@@ -34606,12 +34803,12 @@ static PyObject *__pyx_pw_3_sa_11SuffixArray_21write_enhanced(PyObject *__pyx_v_
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_r = __pyx_pf_3_sa_11SuffixArray_20write_enhanced(((struct __pyx_obj_3_sa_SuffixArray *)__pyx_v_self), ((char *)__pyx_v_filename));
+  __pyx_r = __pyx_pf_3_sa_11SuffixArray_14write_enhanced(((struct __pyx_obj_3_sa_SuffixArray *)__pyx_v_self), ((char *)__pyx_v_filename));
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":197
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":188
  *         fclose(f)
  * 
  *     def write_enhanced(self, char* filename):             # <<<<<<<<<<<<<<
@@ -34619,7 +34816,7 @@ static PyObject *__pyx_pw_3_sa_11SuffixArray_21write_enhanced(PyObject *__pyx_v_
  *             self.darray.write_enhanced_handle(f)
  */
 
-static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, char *__pyx_v_filename) {
+static PyObject *__pyx_pf_3_sa_11SuffixArray_14write_enhanced(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, char *__pyx_v_filename) {
   PyObject *__pyx_v_f = NULL;
   PyObject *__pyx_v_a_i = NULL;
   PyObject *__pyx_v_w_i = NULL;
@@ -34643,7 +34840,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("write_enhanced", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":198
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":189
  * 
  *     def write_enhanced(self, char* filename):
  *         with open(filename, "w") as f:             # <<<<<<<<<<<<<<
@@ -34651,9 +34848,9 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3
  *             for a_i in self.sa:
  */
   /*with:*/ {
-    __pyx_t_1 = PyBytes_FromString(__pyx_v_filename); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 198; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = PyBytes_FromString(__pyx_v_filename); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 189; __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[12]; __pyx_lineno = 198; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 189; __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));
@@ -34661,14 +34858,14 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3
     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[12]; __pyx_lineno = 198; __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[12]; __pyx_lineno = 189; __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[12]; __pyx_lineno = 198; __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[12]; __pyx_lineno = 189; __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[12]; __pyx_lineno = 198; __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[12]; __pyx_lineno = 189; __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[12]; __pyx_lineno = 198; __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[12]; __pyx_lineno = 189; __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;
@@ -34683,27 +34880,27 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3
           __pyx_v_f = __pyx_t_4;
           __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":199
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":190
  *     def write_enhanced(self, char* filename):
  *         with open(filename, "w") as f:
  *             self.darray.write_enhanced_handle(f)             # <<<<<<<<<<<<<<
  *             for a_i in self.sa:
  *                 f.write("%d " % a_i)
  */
-          __pyx_t_4 = PyObject_GetAttr(((PyObject *)__pyx_v_self->darray), __pyx_n_s_27); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 199; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+          __pyx_t_4 = PyObject_GetAttr(((PyObject *)__pyx_v_self->darray), __pyx_n_s_27); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 190; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
           __Pyx_GOTREF(__pyx_t_4);
-          __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 199; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+          __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 190; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
           __Pyx_GOTREF(__pyx_t_1);
           __Pyx_INCREF(__pyx_v_f);
           PyTuple_SET_ITEM(__pyx_t_1, 0, __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[12]; __pyx_lineno = 199; __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[12]; __pyx_lineno = 190; __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;
           __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":200
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":191
  *         with open(filename, "w") as f:
  *             self.darray.write_enhanced_handle(f)
  *             for a_i in self.sa:             # <<<<<<<<<<<<<<
@@ -34714,7 +34911,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3
             __pyx_t_2 = ((PyObject *)__pyx_v_self->sa); __Pyx_INCREF(__pyx_t_2); __pyx_t_8 = 0;
             __pyx_t_9 = NULL;
           } else {
-            __pyx_t_8 = -1; __pyx_t_2 = PyObject_GetIter(((PyObject *)__pyx_v_self->sa)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 200; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+            __pyx_t_8 = -1; __pyx_t_2 = PyObject_GetIter(((PyObject *)__pyx_v_self->sa)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 191; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
             __Pyx_GOTREF(__pyx_t_2);
             __pyx_t_9 = Py_TYPE(__pyx_t_2)->tp_iternext;
           }
@@ -34722,23 +34919,23 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3
             if (!__pyx_t_9 && PyList_CheckExact(__pyx_t_2)) {
               if (__pyx_t_8 >= PyList_GET_SIZE(__pyx_t_2)) break;
               #if CYTHON_COMPILING_IN_CPYTHON
-              __pyx_t_1 = PyList_GET_ITEM(__pyx_t_2, __pyx_t_8); __Pyx_INCREF(__pyx_t_1); __pyx_t_8++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 200; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_1 = PyList_GET_ITEM(__pyx_t_2, __pyx_t_8); __Pyx_INCREF(__pyx_t_1); __pyx_t_8++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 191; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               #else
-              __pyx_t_1 = PySequence_ITEM(__pyx_t_2, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 200; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_1 = PySequence_ITEM(__pyx_t_2, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 191; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               #endif
             } else if (!__pyx_t_9 && PyTuple_CheckExact(__pyx_t_2)) {
               if (__pyx_t_8 >= PyTuple_GET_SIZE(__pyx_t_2)) break;
               #if CYTHON_COMPILING_IN_CPYTHON
-              __pyx_t_1 = PyTuple_GET_ITEM(__pyx_t_2, __pyx_t_8); __Pyx_INCREF(__pyx_t_1); __pyx_t_8++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 200; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_1 = PyTuple_GET_ITEM(__pyx_t_2, __pyx_t_8); __Pyx_INCREF(__pyx_t_1); __pyx_t_8++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 191; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               #else
-              __pyx_t_1 = PySequence_ITEM(__pyx_t_2, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 200; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_1 = PySequence_ITEM(__pyx_t_2, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 191; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               #endif
             } else {
               __pyx_t_1 = __pyx_t_9(__pyx_t_2);
               if (unlikely(!__pyx_t_1)) {
                 if (PyErr_Occurred()) {
                   if (likely(PyErr_ExceptionMatches(PyExc_StopIteration))) PyErr_Clear();
-                  else {__pyx_filename = __pyx_f[12]; __pyx_lineno = 200; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+                  else {__pyx_filename = __pyx_f[12]; __pyx_lineno = 191; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
                 }
                 break;
               }
@@ -34748,23 +34945,23 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3
             __pyx_v_a_i = __pyx_t_1;
             __pyx_t_1 = 0;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":201
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":192
  *             self.darray.write_enhanced_handle(f)
  *             for a_i in self.sa:
  *                 f.write("%d " % a_i)             # <<<<<<<<<<<<<<
  *             f.write("\n")
  *             for w_i in self.ha:
  */
-            __pyx_t_1 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 201; __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[12]; __pyx_lineno = 192; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
             __Pyx_GOTREF(__pyx_t_1);
-            __pyx_t_4 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_21), __pyx_v_a_i); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 201; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+            __pyx_t_4 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_21), __pyx_v_a_i); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 192; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
             __Pyx_GOTREF(((PyObject *)__pyx_t_4));
-            __pyx_t_10 = PyTuple_New(1); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 201; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+            __pyx_t_10 = PyTuple_New(1); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 192; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
             __Pyx_GOTREF(__pyx_t_10);
             PyTuple_SET_ITEM(__pyx_t_10, 0, ((PyObject *)__pyx_t_4));
             __Pyx_GIVEREF(((PyObject *)__pyx_t_4));
             __pyx_t_4 = 0;
-            __pyx_t_4 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_10), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 201; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+            __pyx_t_4 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_10), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 192; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
             __Pyx_GOTREF(__pyx_t_4);
             __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
             __Pyx_DECREF(((PyObject *)__pyx_t_10)); __pyx_t_10 = 0;
@@ -34772,21 +34969,21 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3
           }
           __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":202
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":193
  *             for a_i in self.sa:
  *                 f.write("%d " % a_i)
  *             f.write("\n")             # <<<<<<<<<<<<<<
  *             for w_i in self.ha:
  *                 f.write("%d " % w_i)
  */
-          __pyx_t_2 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 202; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+          __pyx_t_2 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 193; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
           __Pyx_GOTREF(__pyx_t_2);
-          __pyx_t_4 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_k_tuple_97), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 202; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+          __pyx_t_4 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_k_tuple_96), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 193; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
           __Pyx_GOTREF(__pyx_t_4);
           __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
           __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":203
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":194
  *                 f.write("%d " % a_i)
  *             f.write("\n")
  *             for w_i in self.ha:             # <<<<<<<<<<<<<<
@@ -34797,7 +34994,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3
             __pyx_t_4 = ((PyObject *)__pyx_v_self->ha); __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->ha)); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 203; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+            __pyx_t_8 = -1; __pyx_t_4 = PyObject_GetIter(((PyObject *)__pyx_v_self->ha)); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 194; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
             __Pyx_GOTREF(__pyx_t_4);
             __pyx_t_9 = Py_TYPE(__pyx_t_4)->tp_iternext;
           }
@@ -34805,23 +35002,23 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3
             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_2 = PyList_GET_ITEM(__pyx_t_4, __pyx_t_8); __Pyx_INCREF(__pyx_t_2); __pyx_t_8++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 203; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_2 = PyList_GET_ITEM(__pyx_t_4, __pyx_t_8); __Pyx_INCREF(__pyx_t_2); __pyx_t_8++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 194; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               #else
-              __pyx_t_2 = PySequence_ITEM(__pyx_t_4, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 203; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_2 = PySequence_ITEM(__pyx_t_4, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 194; __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_2 = PyTuple_GET_ITEM(__pyx_t_4, __pyx_t_8); __Pyx_INCREF(__pyx_t_2); __pyx_t_8++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 203; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_2 = PyTuple_GET_ITEM(__pyx_t_4, __pyx_t_8); __Pyx_INCREF(__pyx_t_2); __pyx_t_8++; if (unlikely(0 < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 194; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               #else
-              __pyx_t_2 = PySequence_ITEM(__pyx_t_4, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 203; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+              __pyx_t_2 = PySequence_ITEM(__pyx_t_4, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 194; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
               #endif
             } else {
               __pyx_t_2 = __pyx_t_9(__pyx_t_4);
               if (unlikely(!__pyx_t_2)) {
                 if (PyErr_Occurred()) {
                   if (likely(PyErr_ExceptionMatches(PyExc_StopIteration))) PyErr_Clear();
-                  else {__pyx_filename = __pyx_f[12]; __pyx_lineno = 203; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+                  else {__pyx_filename = __pyx_f[12]; __pyx_lineno = 194; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
                 }
                 break;
               }
@@ -34831,23 +35028,23 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3
             __pyx_v_w_i = __pyx_t_2;
             __pyx_t_2 = 0;
 
-            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":204
+            /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":195
  *             f.write("\n")
  *             for w_i in self.ha:
  *                 f.write("%d " % w_i)             # <<<<<<<<<<<<<<
  *             f.write("\n")
  * 
  */
-            __pyx_t_2 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 204; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+            __pyx_t_2 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 195; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
             __Pyx_GOTREF(__pyx_t_2);
-            __pyx_t_10 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_21), __pyx_v_w_i); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 204; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+            __pyx_t_10 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_21), __pyx_v_w_i); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 195; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
             __Pyx_GOTREF(((PyObject *)__pyx_t_10));
-            __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 204; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+            __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 195; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
             __Pyx_GOTREF(__pyx_t_1);
             PyTuple_SET_ITEM(__pyx_t_1, 0, ((PyObject *)__pyx_t_10));
             __Pyx_GIVEREF(((PyObject *)__pyx_t_10));
             __pyx_t_10 = 0;
-            __pyx_t_10 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 204; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+            __pyx_t_10 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 195; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
             __Pyx_GOTREF(__pyx_t_10);
             __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
             __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
@@ -34855,16 +35052,16 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3
           }
           __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":205
+          /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":196
  *             for w_i in self.ha:
  *                 f.write("%d " % w_i)
  *             f.write("\n")             # <<<<<<<<<<<<<<
  * 
  *     cdef int __search_high(self, int word_id, int offset, int low, int high):
  */
-          __pyx_t_4 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 205; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+          __pyx_t_4 = PyObject_GetAttr(__pyx_v_f, __pyx_n_s__write); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 196; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
           __Pyx_GOTREF(__pyx_t_4);
-          __pyx_t_10 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_k_tuple_98), NULL); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 205; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+          __pyx_t_10 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_k_tuple_97), NULL); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 196; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
           __Pyx_GOTREF(__pyx_t_10);
           __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
           __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
@@ -34879,7 +35076,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3
         __Pyx_XDECREF(__pyx_t_4); __pyx_t_4 = 0;
         __Pyx_XDECREF(__pyx_t_10); __pyx_t_10 = 0;
 
-        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":198
+        /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":189
  * 
  *     def write_enhanced(self, char* filename):
  *         with open(filename, "w") as f:             # <<<<<<<<<<<<<<
@@ -34888,11 +35085,11 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3
  */
         /*except:*/ {
           __Pyx_AddTraceback("_sa.SuffixArray.write_enhanced", __pyx_clineno, __pyx_lineno, __pyx_filename);
-          if (__Pyx_GetException(&__pyx_t_10, &__pyx_t_4, &__pyx_t_1) < 0) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 198; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (__Pyx_GetException(&__pyx_t_10, &__pyx_t_4, &__pyx_t_1) < 0) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 189; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __Pyx_GOTREF(__pyx_t_10);
           __Pyx_GOTREF(__pyx_t_4);
           __Pyx_GOTREF(__pyx_t_1);
-          __pyx_t_2 = PyTuple_New(3); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 198; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          __pyx_t_2 = PyTuple_New(3); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 189; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __Pyx_GOTREF(__pyx_t_2);
           __Pyx_INCREF(__pyx_t_10);
           PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_t_10);
@@ -34905,11 +35102,11 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3
           __Pyx_GIVEREF(__pyx_t_1);
           __pyx_t_12 = PyObject_Call(__pyx_t_3, __pyx_t_2, NULL);
           __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-          if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 198; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (unlikely(!__pyx_t_12)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 189; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __Pyx_GOTREF(__pyx_t_12);
           __pyx_t_11 = __Pyx_PyObject_IsTrue(__pyx_t_12);
           __Pyx_DECREF(__pyx_t_12); __pyx_t_12 = 0;
-          if (unlikely(__pyx_t_11 < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 198; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+          if (unlikely(__pyx_t_11 < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 189; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
           __pyx_t_13 = (!__pyx_t_11);
           if (__pyx_t_13) {
             __Pyx_GIVEREF(__pyx_t_10);
@@ -34917,7 +35114,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3
             __Pyx_GIVEREF(__pyx_t_1);
             __Pyx_ErrRestore(__pyx_t_10, __pyx_t_4, __pyx_t_1);
             __pyx_t_10 = 0; __pyx_t_4 = 0; __pyx_t_1 = 0; 
-            {__pyx_filename = __pyx_f[12]; __pyx_lineno = 198; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+            {__pyx_filename = __pyx_f[12]; __pyx_lineno = 189; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
             goto __pyx_L22;
           }
           __pyx_L22:;
@@ -34943,13 +35140,13 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3
     }
     /*finally:*/ {
       if (__pyx_t_3) {
-        __pyx_t_7 = PyObject_Call(__pyx_t_3, __pyx_k_tuple_99, NULL);
+        __pyx_t_7 = PyObject_Call(__pyx_t_3, __pyx_k_tuple_98, NULL);
         __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-        if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 198; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 189; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
         __Pyx_GOTREF(__pyx_t_7);
         __pyx_t_13 = __Pyx_PyObject_IsTrue(__pyx_t_7);
         __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
-        if (unlikely(__pyx_t_13 < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 198; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        if (unlikely(__pyx_t_13 < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 189; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       }
     }
     goto __pyx_L23;
@@ -34977,7 +35174,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_20write_enhanced(struct __pyx_obj_3
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":207
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":198
  *             f.write("\n")
  * 
  *     cdef int __search_high(self, int word_id, int offset, int low, int high):             # <<<<<<<<<<<<<<
@@ -34992,7 +35189,7 @@ static int __pyx_f_3_sa_11SuffixArray___search_high(struct __pyx_obj_3_sa_Suffix
   int __pyx_t_1;
   __Pyx_RefNannySetupContext("__search_high", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":210
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":201
  *         cdef int midpoint
  * 
  *         if low >= high:             # <<<<<<<<<<<<<<
@@ -35002,7 +35199,7 @@ static int __pyx_f_3_sa_11SuffixArray___search_high(struct __pyx_obj_3_sa_Suffix
   __pyx_t_1 = (__pyx_v_low >= __pyx_v_high);
   if (__pyx_t_1) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":211
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":202
  * 
  *         if low >= high:
  *             return high             # <<<<<<<<<<<<<<
@@ -35015,7 +35212,7 @@ static int __pyx_f_3_sa_11SuffixArray___search_high(struct __pyx_obj_3_sa_Suffix
   }
   __pyx_L3:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":212
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":203
  *         if low >= high:
  *             return high
  *         midpoint = (high + low) / 2             # <<<<<<<<<<<<<<
@@ -35024,7 +35221,7 @@ static int __pyx_f_3_sa_11SuffixArray___search_high(struct __pyx_obj_3_sa_Suffix
  */
   __pyx_v_midpoint = __Pyx_div_long((__pyx_v_high + __pyx_v_low), 2);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":213
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":204
  *             return high
  *         midpoint = (high + low) / 2
  *         if self.darray.data.arr[self.sa.arr[midpoint] + offset] == word_id:             # <<<<<<<<<<<<<<
@@ -35034,7 +35231,7 @@ static int __pyx_f_3_sa_11SuffixArray___search_high(struct __pyx_obj_3_sa_Suffix
   __pyx_t_1 = ((__pyx_v_self->darray->data->arr[((__pyx_v_self->sa->arr[__pyx_v_midpoint]) + __pyx_v_offset)]) == __pyx_v_word_id);
   if (__pyx_t_1) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":214
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":205
  *         midpoint = (high + low) / 2
  *         if self.darray.data.arr[self.sa.arr[midpoint] + offset] == word_id:
  *             return self.__search_high(word_id, offset, midpoint+1, high)             # <<<<<<<<<<<<<<
@@ -35047,7 +35244,7 @@ static int __pyx_f_3_sa_11SuffixArray___search_high(struct __pyx_obj_3_sa_Suffix
   }
   /*else*/ {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":216
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":207
  *             return self.__search_high(word_id, offset, midpoint+1, high)
  *         else:
  *             return self.__search_high(word_id, offset, low, midpoint)             # <<<<<<<<<<<<<<
@@ -35065,7 +35262,7 @@ static int __pyx_f_3_sa_11SuffixArray___search_high(struct __pyx_obj_3_sa_Suffix
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":218
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":209
  *             return self.__search_high(word_id, offset, low, midpoint)
  * 
  *     cdef int __search_low(self, int word_id, int offset, int low, int high):             # <<<<<<<<<<<<<<
@@ -35080,7 +35277,7 @@ static int __pyx_f_3_sa_11SuffixArray___search_low(struct __pyx_obj_3_sa_SuffixA
   int __pyx_t_1;
   __Pyx_RefNannySetupContext("__search_low", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":221
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":212
  *         cdef int midpoint
  * 
  *         if low >= high:             # <<<<<<<<<<<<<<
@@ -35090,7 +35287,7 @@ static int __pyx_f_3_sa_11SuffixArray___search_low(struct __pyx_obj_3_sa_SuffixA
   __pyx_t_1 = (__pyx_v_low >= __pyx_v_high);
   if (__pyx_t_1) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":222
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":213
  * 
  *         if low >= high:
  *             return high             # <<<<<<<<<<<<<<
@@ -35103,7 +35300,7 @@ static int __pyx_f_3_sa_11SuffixArray___search_low(struct __pyx_obj_3_sa_SuffixA
   }
   __pyx_L3:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":223
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":214
  *         if low >= high:
  *             return high
  *         midpoint = (high + low) / 2             # <<<<<<<<<<<<<<
@@ -35112,7 +35309,7 @@ static int __pyx_f_3_sa_11SuffixArray___search_low(struct __pyx_obj_3_sa_SuffixA
  */
   __pyx_v_midpoint = __Pyx_div_long((__pyx_v_high + __pyx_v_low), 2);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":224
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":215
  *             return high
  *         midpoint = (high + low) / 2
  *         if self.darray.data.arr[self.sa.arr[midpoint] + offset] == word_id:             # <<<<<<<<<<<<<<
@@ -35122,7 +35319,7 @@ static int __pyx_f_3_sa_11SuffixArray___search_low(struct __pyx_obj_3_sa_SuffixA
   __pyx_t_1 = ((__pyx_v_self->darray->data->arr[((__pyx_v_self->sa->arr[__pyx_v_midpoint]) + __pyx_v_offset)]) == __pyx_v_word_id);
   if (__pyx_t_1) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":225
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":216
  *         midpoint = (high + low) / 2
  *         if self.darray.data.arr[self.sa.arr[midpoint] + offset] == word_id:
  *             return self.__search_low(word_id, offset, low, midpoint)             # <<<<<<<<<<<<<<
@@ -35135,7 +35332,7 @@ static int __pyx_f_3_sa_11SuffixArray___search_low(struct __pyx_obj_3_sa_SuffixA
   }
   /*else*/ {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":227
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":218
  *             return self.__search_low(word_id, offset, low, midpoint)
  *         else:
  *             return self.__search_low(word_id, offset, midpoint+1, high)             # <<<<<<<<<<<<<<
@@ -35153,7 +35350,7 @@ static int __pyx_f_3_sa_11SuffixArray___search_low(struct __pyx_obj_3_sa_SuffixA
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":229
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":220
  *             return self.__search_low(word_id, offset, midpoint+1, high)
  * 
  *     cdef __get_range(self, int word_id, int offset, int low, int high, int midpoint):             # <<<<<<<<<<<<<<
@@ -35172,7 +35369,7 @@ static PyObject *__pyx_f_3_sa_11SuffixArray___get_range(struct __pyx_obj_3_sa_Su
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("__get_range", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":230
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":221
  * 
  *     cdef __get_range(self, int word_id, int offset, int low, int high, int midpoint):
  *         return (self.__search_low(word_id, offset, low, midpoint),             # <<<<<<<<<<<<<<
@@ -35180,19 +35377,19 @@ static PyObject *__pyx_f_3_sa_11SuffixArray___get_range(struct __pyx_obj_3_sa_Su
  * 
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_1 = PyInt_FromLong(((struct __pyx_vtabstruct_3_sa_SuffixArray *)__pyx_v_self->__pyx_vtab)->__pyx___search_low(__pyx_v_self, __pyx_v_word_id, __pyx_v_offset, __pyx_v_low, __pyx_v_midpoint)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 230; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyInt_FromLong(((struct __pyx_vtabstruct_3_sa_SuffixArray *)__pyx_v_self->__pyx_vtab)->__pyx___search_low(__pyx_v_self, __pyx_v_word_id, __pyx_v_offset, __pyx_v_low, __pyx_v_midpoint)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 221; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":231
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":222
  *     cdef __get_range(self, int word_id, int offset, int low, int high, int midpoint):
  *         return (self.__search_low(word_id, offset, low, midpoint),
  *                 self.__search_high(word_id, offset, midpoint, high))             # <<<<<<<<<<<<<<
  * 
  *     cdef __lookup_helper(self, int word_id, int offset, int low, int high):
  */
-  __pyx_t_2 = PyInt_FromLong(((struct __pyx_vtabstruct_3_sa_SuffixArray *)__pyx_v_self->__pyx_vtab)->__pyx___search_high(__pyx_v_self, __pyx_v_word_id, __pyx_v_offset, __pyx_v_midpoint, __pyx_v_high)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 231; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyInt_FromLong(((struct __pyx_vtabstruct_3_sa_SuffixArray *)__pyx_v_self->__pyx_vtab)->__pyx___search_high(__pyx_v_self, __pyx_v_word_id, __pyx_v_offset, __pyx_v_midpoint, __pyx_v_high)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 222; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyTuple_New(2); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 230; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyTuple_New(2); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 221; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   PyTuple_SET_ITEM(__pyx_t_3, 0, __pyx_t_1);
   __Pyx_GIVEREF(__pyx_t_1);
@@ -35218,7 +35415,7 @@ static PyObject *__pyx_f_3_sa_11SuffixArray___get_range(struct __pyx_obj_3_sa_Su
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":233
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":224
  *                 self.__search_high(word_id, offset, midpoint, high))
  * 
  *     cdef __lookup_helper(self, int word_id, int offset, int low, int high):             # <<<<<<<<<<<<<<
@@ -35239,7 +35436,7 @@ static PyObject *__pyx_f_3_sa_11SuffixArray___lookup_helper(struct __pyx_obj_3_s
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("__lookup_helper", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":236
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":227
  *         cdef int midpoint
  * 
  *         if offset == 0:             # <<<<<<<<<<<<<<
@@ -35249,7 +35446,7 @@ static PyObject *__pyx_f_3_sa_11SuffixArray___lookup_helper(struct __pyx_obj_3_s
   __pyx_t_1 = (__pyx_v_offset == 0);
   if (__pyx_t_1) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":237
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":228
  * 
  *         if offset == 0:
  *             return (self.ha.arr[word_id], self.ha.arr[word_id+1])             # <<<<<<<<<<<<<<
@@ -35257,11 +35454,11 @@ static PyObject *__pyx_f_3_sa_11SuffixArray___lookup_helper(struct __pyx_obj_3_s
  *             return None
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = PyInt_FromLong((__pyx_v_self->ha->arr[__pyx_v_word_id])); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 237; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyInt_FromLong((__pyx_v_self->ha->arr[__pyx_v_word_id])); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 228; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_3 = PyInt_FromLong((__pyx_v_self->ha->arr[(__pyx_v_word_id + 1)])); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 237; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyInt_FromLong((__pyx_v_self->ha->arr[(__pyx_v_word_id + 1)])); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 228; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
-    __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 237; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 228; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
     PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_2);
     __Pyx_GIVEREF(__pyx_t_2);
@@ -35276,7 +35473,7 @@ static PyObject *__pyx_f_3_sa_11SuffixArray___lookup_helper(struct __pyx_obj_3_s
   }
   __pyx_L3:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":238
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":229
  *         if offset == 0:
  *             return (self.ha.arr[word_id], self.ha.arr[word_id+1])
  *         if low >= high:             # <<<<<<<<<<<<<<
@@ -35286,7 +35483,7 @@ static PyObject *__pyx_f_3_sa_11SuffixArray___lookup_helper(struct __pyx_obj_3_s
   __pyx_t_1 = (__pyx_v_low >= __pyx_v_high);
   if (__pyx_t_1) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":239
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":230
  *             return (self.ha.arr[word_id], self.ha.arr[word_id+1])
  *         if low >= high:
  *             return None             # <<<<<<<<<<<<<<
@@ -35301,7 +35498,7 @@ static PyObject *__pyx_f_3_sa_11SuffixArray___lookup_helper(struct __pyx_obj_3_s
   }
   __pyx_L4:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":241
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":232
  *             return None
  * 
  *         midpoint = (high + low) / 2             # <<<<<<<<<<<<<<
@@ -35310,7 +35507,7 @@ static PyObject *__pyx_f_3_sa_11SuffixArray___lookup_helper(struct __pyx_obj_3_s
  */
   __pyx_v_midpoint = __Pyx_div_long((__pyx_v_high + __pyx_v_low), 2);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":242
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":233
  * 
  *         midpoint = (high + low) / 2
  *         if self.darray.data.arr[self.sa.arr[midpoint] + offset] == word_id:             # <<<<<<<<<<<<<<
@@ -35320,7 +35517,7 @@ static PyObject *__pyx_f_3_sa_11SuffixArray___lookup_helper(struct __pyx_obj_3_s
   __pyx_t_1 = ((__pyx_v_self->darray->data->arr[((__pyx_v_self->sa->arr[__pyx_v_midpoint]) + __pyx_v_offset)]) == __pyx_v_word_id);
   if (__pyx_t_1) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":243
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":234
  *         midpoint = (high + low) / 2
  *         if self.darray.data.arr[self.sa.arr[midpoint] + offset] == word_id:
  *             return self.__get_range(word_id, offset, low, high, midpoint)             # <<<<<<<<<<<<<<
@@ -35328,7 +35525,7 @@ static PyObject *__pyx_f_3_sa_11SuffixArray___lookup_helper(struct __pyx_obj_3_s
  *             return self.__lookup_helper(word_id, offset, low, midpoint)
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_4 = ((struct __pyx_vtabstruct_3_sa_SuffixArray *)__pyx_v_self->__pyx_vtab)->__pyx___get_range(__pyx_v_self, __pyx_v_word_id, __pyx_v_offset, __pyx_v_low, __pyx_v_high, __pyx_v_midpoint); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 243; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = ((struct __pyx_vtabstruct_3_sa_SuffixArray *)__pyx_v_self->__pyx_vtab)->__pyx___get_range(__pyx_v_self, __pyx_v_word_id, __pyx_v_offset, __pyx_v_low, __pyx_v_high, __pyx_v_midpoint); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 234; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
     __pyx_r = __pyx_t_4;
     __pyx_t_4 = 0;
@@ -35337,7 +35534,7 @@ static PyObject *__pyx_f_3_sa_11SuffixArray___lookup_helper(struct __pyx_obj_3_s
   }
   __pyx_L5:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":244
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":235
  *         if self.darray.data.arr[self.sa.arr[midpoint] + offset] == word_id:
  *             return self.__get_range(word_id, offset, low, high, midpoint)
  *         if self.darray.data.arr[self.sa.arr[midpoint] + offset] > word_id:             # <<<<<<<<<<<<<<
@@ -35347,7 +35544,7 @@ static PyObject *__pyx_f_3_sa_11SuffixArray___lookup_helper(struct __pyx_obj_3_s
   __pyx_t_1 = ((__pyx_v_self->darray->data->arr[((__pyx_v_self->sa->arr[__pyx_v_midpoint]) + __pyx_v_offset)]) > __pyx_v_word_id);
   if (__pyx_t_1) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":245
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":236
  *             return self.__get_range(word_id, offset, low, high, midpoint)
  *         if self.darray.data.arr[self.sa.arr[midpoint] + offset] > word_id:
  *             return self.__lookup_helper(word_id, offset, low, midpoint)             # <<<<<<<<<<<<<<
@@ -35355,7 +35552,7 @@ static PyObject *__pyx_f_3_sa_11SuffixArray___lookup_helper(struct __pyx_obj_3_s
  *             return self.__lookup_helper(word_id, offset, midpoint+1, high)
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_4 = ((struct __pyx_vtabstruct_3_sa_SuffixArray *)__pyx_v_self->__pyx_vtab)->__pyx___lookup_helper(__pyx_v_self, __pyx_v_word_id, __pyx_v_offset, __pyx_v_low, __pyx_v_midpoint); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 245; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = ((struct __pyx_vtabstruct_3_sa_SuffixArray *)__pyx_v_self->__pyx_vtab)->__pyx___lookup_helper(__pyx_v_self, __pyx_v_word_id, __pyx_v_offset, __pyx_v_low, __pyx_v_midpoint); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 236; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
     __pyx_r = __pyx_t_4;
     __pyx_t_4 = 0;
@@ -35364,7 +35561,7 @@ static PyObject *__pyx_f_3_sa_11SuffixArray___lookup_helper(struct __pyx_obj_3_s
   }
   /*else*/ {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":247
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":238
  *             return self.__lookup_helper(word_id, offset, low, midpoint)
  *         else:
  *             return self.__lookup_helper(word_id, offset, midpoint+1, high)             # <<<<<<<<<<<<<<
@@ -35372,7 +35569,7 @@ static PyObject *__pyx_f_3_sa_11SuffixArray___lookup_helper(struct __pyx_obj_3_s
  *     def lookup(self, word, int offset, int low, int high):
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_4 = ((struct __pyx_vtabstruct_3_sa_SuffixArray *)__pyx_v_self->__pyx_vtab)->__pyx___lookup_helper(__pyx_v_self, __pyx_v_word_id, __pyx_v_offset, (__pyx_v_midpoint + 1), __pyx_v_high); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 247; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = ((struct __pyx_vtabstruct_3_sa_SuffixArray *)__pyx_v_self->__pyx_vtab)->__pyx___lookup_helper(__pyx_v_self, __pyx_v_word_id, __pyx_v_offset, (__pyx_v_midpoint + 1), __pyx_v_high); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 238; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
     __pyx_r = __pyx_t_4;
     __pyx_t_4 = 0;
@@ -35395,8 +35592,8 @@ static PyObject *__pyx_f_3_sa_11SuffixArray___lookup_helper(struct __pyx_obj_3_s
 }
 
 /* Python wrapper */
-static PyObject *__pyx_pw_3_sa_11SuffixArray_23lookup(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static PyObject *__pyx_pw_3_sa_11SuffixArray_23lookup(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pw_3_sa_11SuffixArray_17lookup(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static PyObject *__pyx_pw_3_sa_11SuffixArray_17lookup(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_word = 0;
   int __pyx_v_offset;
   int __pyx_v_low;
@@ -35426,21 +35623,21 @@ static PyObject *__pyx_pw_3_sa_11SuffixArray_23lookup(PyObject *__pyx_v_self, Py
         case  1:
         if (likely((values[1] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__offset)) != 0)) kw_args--;
         else {
-          __Pyx_RaiseArgtupleInvalid("lookup", 1, 4, 4, 1); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 249; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+          __Pyx_RaiseArgtupleInvalid("lookup", 1, 4, 4, 1); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 240; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
         }
         case  2:
         if (likely((values[2] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__low)) != 0)) kw_args--;
         else {
-          __Pyx_RaiseArgtupleInvalid("lookup", 1, 4, 4, 2); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 249; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+          __Pyx_RaiseArgtupleInvalid("lookup", 1, 4, 4, 2); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 240; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
         }
         case  3:
         if (likely((values[3] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__high)) != 0)) kw_args--;
         else {
-          __Pyx_RaiseArgtupleInvalid("lookup", 1, 4, 4, 3); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 249; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+          __Pyx_RaiseArgtupleInvalid("lookup", 1, 4, 4, 3); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 240; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
         }
       }
       if (unlikely(kw_args > 0)) {
-        if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "lookup") < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 249; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+        if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "lookup") < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 240; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
     } else if (PyTuple_GET_SIZE(__pyx_args) != 4) {
       goto __pyx_L5_argtuple_error;
@@ -35451,24 +35648,24 @@ static PyObject *__pyx_pw_3_sa_11SuffixArray_23lookup(PyObject *__pyx_v_self, Py
       values[3] = PyTuple_GET_ITEM(__pyx_args, 3);
     }
     __pyx_v_word = values[0];
-    __pyx_v_offset = __Pyx_PyInt_AsInt(values[1]); if (unlikely((__pyx_v_offset == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 249; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
-    __pyx_v_low = __Pyx_PyInt_AsInt(values[2]); if (unlikely((__pyx_v_low == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 249; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
-    __pyx_v_high = __Pyx_PyInt_AsInt(values[3]); if (unlikely((__pyx_v_high == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 249; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_v_offset = __Pyx_PyInt_AsInt(values[1]); if (unlikely((__pyx_v_offset == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 240; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_v_low = __Pyx_PyInt_AsInt(values[2]); if (unlikely((__pyx_v_low == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 240; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_v_high = __Pyx_PyInt_AsInt(values[3]); if (unlikely((__pyx_v_high == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 240; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("lookup", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 249; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("lookup", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[12]; __pyx_lineno = 240; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
   __Pyx_AddTraceback("_sa.SuffixArray.lookup", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_r = __pyx_pf_3_sa_11SuffixArray_22lookup(((struct __pyx_obj_3_sa_SuffixArray *)__pyx_v_self), __pyx_v_word, __pyx_v_offset, __pyx_v_low, __pyx_v_high);
+  __pyx_r = __pyx_pf_3_sa_11SuffixArray_16lookup(((struct __pyx_obj_3_sa_SuffixArray *)__pyx_v_self), __pyx_v_word, __pyx_v_offset, __pyx_v_low, __pyx_v_high);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":249
+/* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":240
  *             return self.__lookup_helper(word_id, offset, midpoint+1, high)
  * 
  *     def lookup(self, word, int offset, int low, int high):             # <<<<<<<<<<<<<<
@@ -35476,7 +35673,7 @@ static PyObject *__pyx_pw_3_sa_11SuffixArray_23lookup(PyObject *__pyx_v_self, Py
  *         if low == -1:
  */
 
-static PyObject *__pyx_pf_3_sa_11SuffixArray_22lookup(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, PyObject *__pyx_v_word, int __pyx_v_offset, int __pyx_v_low, int __pyx_v_high) {
+static PyObject *__pyx_pf_3_sa_11SuffixArray_16lookup(struct __pyx_obj_3_sa_SuffixArray *__pyx_v_self, PyObject *__pyx_v_word, int __pyx_v_offset, int __pyx_v_low, int __pyx_v_high) {
   PyObject *__pyx_v_word_id = NULL;
   PyObject *__pyx_r = NULL;
   __Pyx_RefNannyDeclarations
@@ -35489,7 +35686,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_22lookup(struct __pyx_obj_3_sa_Suff
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("lookup", 0);
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":251
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":242
  *     def lookup(self, word, int offset, int low, int high):
  *         cdef int wordid
  *         if low == -1:             # <<<<<<<<<<<<<<
@@ -35499,7 +35696,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_22lookup(struct __pyx_obj_3_sa_Suff
   __pyx_t_1 = (__pyx_v_low == -1);
   if (__pyx_t_1) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":252
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":243
  *         cdef int wordid
  *         if low == -1:
  *             low = 0             # <<<<<<<<<<<<<<
@@ -35511,7 +35708,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_22lookup(struct __pyx_obj_3_sa_Suff
   }
   __pyx_L3:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":253
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":244
  *         if low == -1:
  *             low = 0
  *         if high == -1:             # <<<<<<<<<<<<<<
@@ -35521,7 +35718,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_22lookup(struct __pyx_obj_3_sa_Suff
   __pyx_t_1 = (__pyx_v_high == -1);
   if (__pyx_t_1) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":254
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":245
  *             low = 0
  *         if high == -1:
  *             high = len(self.sa)             # <<<<<<<<<<<<<<
@@ -35530,36 +35727,36 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_22lookup(struct __pyx_obj_3_sa_Suff
  */
     __pyx_t_2 = ((PyObject *)__pyx_v_self->sa);
     __Pyx_INCREF(__pyx_t_2);
-    __pyx_t_3 = PyObject_Length(__pyx_t_2); if (unlikely(__pyx_t_3 == -1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 254; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyObject_Length(__pyx_t_2); if (unlikely(__pyx_t_3 == -1)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 245; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
     __pyx_v_high = __pyx_t_3;
     goto __pyx_L4;
   }
   __pyx_L4:;
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":255
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":246
  *         if high == -1:
  *             high = len(self.sa)
  *         if word in self.darray.word2id:             # <<<<<<<<<<<<<<
  *             word_id = self.darray.word2id[word]
  *             return self.__lookup_helper(word_id, offset, low, high)
  */
-  __pyx_t_1 = (__Pyx_PySequence_Contains(__pyx_v_word, __pyx_v_self->darray->word2id, Py_EQ)); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 255; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = (__Pyx_PySequence_Contains(__pyx_v_word, __pyx_v_self->darray->word2id, Py_EQ)); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 246; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   if (__pyx_t_1) {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":256
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":247
  *             high = len(self.sa)
  *         if word in self.darray.word2id:
  *             word_id = self.darray.word2id[word]             # <<<<<<<<<<<<<<
  *             return self.__lookup_helper(word_id, offset, low, high)
  *         else:
  */
-    __pyx_t_2 = PyObject_GetItem(__pyx_v_self->darray->word2id, __pyx_v_word); if (!__pyx_t_2) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 256; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_GetItem(__pyx_v_self->darray->word2id, __pyx_v_word); if (!__pyx_t_2) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 247; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_v_word_id = __pyx_t_2;
     __pyx_t_2 = 0;
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":257
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":248
  *         if word in self.darray.word2id:
  *             word_id = self.darray.word2id[word]
  *             return self.__lookup_helper(word_id, offset, low, high)             # <<<<<<<<<<<<<<
@@ -35567,8 +35764,8 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_22lookup(struct __pyx_obj_3_sa_Suff
  *             return None
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_4 = __Pyx_PyInt_AsInt(__pyx_v_word_id); if (unlikely((__pyx_t_4 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 257; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-    __pyx_t_2 = ((struct __pyx_vtabstruct_3_sa_SuffixArray *)__pyx_v_self->__pyx_vtab)->__pyx___lookup_helper(__pyx_v_self, __pyx_t_4, __pyx_v_offset, __pyx_v_low, __pyx_v_high); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 257; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = __Pyx_PyInt_AsInt(__pyx_v_word_id); if (unlikely((__pyx_t_4 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 248; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = ((struct __pyx_vtabstruct_3_sa_SuffixArray *)__pyx_v_self->__pyx_vtab)->__pyx___lookup_helper(__pyx_v_self, __pyx_t_4, __pyx_v_offset, __pyx_v_low, __pyx_v_high); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 248; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -35577,7 +35774,7 @@ static PyObject *__pyx_pf_3_sa_11SuffixArray_22lookup(struct __pyx_obj_3_sa_Suff
   }
   /*else*/ {
 
-    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":259
+    /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":250
  *             return self.__lookup_helper(word_id, offset, low, high)
  *         else:
  *             return None             # <<<<<<<<<<<<<<
@@ -36146,7 +36343,7 @@ static int __pyx_pw_3_sa_9TrieTable_1__cinit__(PyObject *__pyx_v_self, PyObject
   {
     static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__extended,0};
     PyObject* values[1] = {0};
-    values[0] = __pyx_k_100;
+    values[0] = __pyx_k_99;
     if (unlikely(__pyx_kwds)) {
       Py_ssize_t kw_args;
       const Py_ssize_t pos_args = PyTuple_GET_SIZE(__pyx_args);
@@ -36893,9 +37090,9 @@ static int __pyx_pf_3_sa_7Sampler___cinit__(struct __pyx_obj_3_sa_Sampler *__pyx
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 105; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
-    __Pyx_INCREF(((PyObject *)__pyx_kp_s_101));
-    PyTuple_SET_ITEM(__pyx_t_4, 0, ((PyObject *)__pyx_kp_s_101));
-    __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_101));
+    __Pyx_INCREF(((PyObject *)__pyx_kp_s_100));
+    PyTuple_SET_ITEM(__pyx_t_4, 0, ((PyObject *)__pyx_kp_s_100));
+    __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_100));
     PyTuple_SET_ITEM(__pyx_t_4, 1, __pyx_t_2);
     __Pyx_GIVEREF(__pyx_t_2);
     __pyx_t_2 = 0;
@@ -36920,7 +37117,7 @@ static int __pyx_pf_3_sa_7Sampler___cinit__(struct __pyx_obj_3_sa_Sampler *__pyx
     __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__info); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_k_tuple_103), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_k_tuple_102), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
@@ -37708,7 +37905,7 @@ static int __pyx_pw_3_sa_23HieroCachingRuleFactory_1__cinit__(PyObject *__pyx_v_
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("__cinit__ (wrapper)", 0);
   {
-    static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__alignment,&__pyx_n_s__by_slack_factor,&__pyx_n_s__category,&__pyx_n_s__max_chunks,&__pyx_n_s__max_initial_size,&__pyx_n_s__max_length,&__pyx_n_s__max_nonterminals,&__pyx_n_s__max_target_chunks,&__pyx_n_s__max_target_length,&__pyx_n_s__min_gap_size,&__pyx_n_s__precompute_file,&__pyx_n_s_70,&__pyx_n_s__precompute_rank,&__pyx_n_s_104,&__pyx_n_s_105,&__pyx_n_s_71,&__pyx_n_s__train_min_gap_size,&__pyx_n_s__tight_phrases,&__pyx_n_s__use_baeza_yates,&__pyx_n_s__use_collocations,&__pyx_n_s__use_index,0};
+    static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__alignment,&__pyx_n_s__by_slack_factor,&__pyx_n_s__category,&__pyx_n_s__max_chunks,&__pyx_n_s__max_initial_size,&__pyx_n_s__max_length,&__pyx_n_s__max_nonterminals,&__pyx_n_s__max_target_chunks,&__pyx_n_s__max_target_length,&__pyx_n_s__min_gap_size,&__pyx_n_s__precompute_file,&__pyx_n_s_70,&__pyx_n_s__precompute_rank,&__pyx_n_s_103,&__pyx_n_s_104,&__pyx_n_s_71,&__pyx_n_s__train_min_gap_size,&__pyx_n_s__tight_phrases,&__pyx_n_s__use_baeza_yates,&__pyx_n_s__use_collocations,&__pyx_n_s__use_index,0};
     PyObject* values[21] = {0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0};
 
     /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":273
@@ -37841,12 +38038,12 @@ static int __pyx_pw_3_sa_23HieroCachingRuleFactory_1__cinit__(PyObject *__pyx_v_
         }
         case 13:
         if (kw_args > 0) {
-          PyObject* value = PyDict_GetItem(__pyx_kwds, __pyx_n_s_104);
+          PyObject* value = PyDict_GetItem(__pyx_kwds, __pyx_n_s_103);
           if (value) { values[13] = value; kw_args--; }
         }
         case 14:
         if (kw_args > 0) {
-          PyObject* value = PyDict_GetItem(__pyx_kwds, __pyx_n_s_105);
+          PyObject* value = PyDict_GetItem(__pyx_kwds, __pyx_n_s_104);
           if (value) { values[14] = value; kw_args--; }
         }
         case 15:
@@ -37927,7 +38124,7 @@ static int __pyx_pw_3_sa_23HieroCachingRuleFactory_1__cinit__(PyObject *__pyx_v_
     if (values[2]) {
       __pyx_v_category = PyBytes_AsString(values[2]); if (unlikely((!__pyx_v_category) && PyErr_Occurred())) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 271; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     } else {
-      __pyx_v_category = ((char *)__pyx_k_106);
+      __pyx_v_category = ((char *)__pyx_k_105);
     }
     __pyx_v_max_chunks = values[3];
     if (values[4]) {
@@ -38152,7 +38349,7 @@ static int __pyx_pf_3_sa_23HieroCachingRuleFactory___cinit__(struct __pyx_obj_3_
  *         self.alignment = alignment
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_Exception, ((PyObject *)__pyx_k_tuple_108), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 316; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_Exception, ((PyObject *)__pyx_k_tuple_107), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 316; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
@@ -39568,9 +39765,9 @@ static PyObject *__pyx_pf_3_sa_23HieroCachingRuleFactory_8precompute(struct __py
     __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 = 440; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_INCREF(((PyObject *)__pyx_kp_s_109));
-    PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_kp_s_109));
-    __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_109));
+    __Pyx_INCREF(((PyObject *)__pyx_kp_s_108));
+    PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_kp_s_108));
+    __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_108));
     __Pyx_INCREF(__pyx_v_self->precompute_file);
     PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_v_self->precompute_file);
     __Pyx_GIVEREF(__pyx_v_self->precompute_file);
@@ -39624,9 +39821,9 @@ static PyObject *__pyx_pf_3_sa_23HieroCachingRuleFactory_8precompute(struct __py
       __Pyx_GOTREF(__pyx_t_3);
       __pyx_t_5 = PyTuple_New(3); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 444; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_5);
-      __Pyx_INCREF(((PyObject *)__pyx_kp_s_110));
-      PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_kp_s_110));
-      __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_110));
+      __Pyx_INCREF(((PyObject *)__pyx_kp_s_109));
+      PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_kp_s_109));
+      __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_109));
       PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
       __Pyx_GIVEREF(__pyx_t_2);
       PyTuple_SET_ITEM(__pyx_t_5, 2, __pyx_t_3);
@@ -39670,9 +39867,9 @@ static PyObject *__pyx_pf_3_sa_23HieroCachingRuleFactory_8precompute(struct __py
       __Pyx_GOTREF(__pyx_t_4);
       __pyx_t_2 = PyTuple_New(3); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 446; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_INCREF(((PyObject *)__pyx_kp_s_111));
-      PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_kp_s_111));
-      __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_111));
+      __Pyx_INCREF(((PyObject *)__pyx_kp_s_110));
+      PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_kp_s_110));
+      __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_110));
       PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_t_3);
       __Pyx_GIVEREF(__pyx_t_3);
       PyTuple_SET_ITEM(__pyx_t_2, 2, __pyx_t_4);
@@ -39717,7 +39914,7 @@ static PyObject *__pyx_pf_3_sa_23HieroCachingRuleFactory_8precompute(struct __py
       __Pyx_GIVEREF(__pyx_t_2);
       __pyx_t_4 = 0;
       __pyx_t_2 = 0;
-      __pyx_t_2 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_112), ((PyObject *)__pyx_t_5)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 448; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_111), ((PyObject *)__pyx_t_5)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 448; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(((PyObject *)__pyx_t_2));
       __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[8]; __pyx_lineno = 448; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
@@ -39764,7 +39961,7 @@ static PyObject *__pyx_pf_3_sa_23HieroCachingRuleFactory_8precompute(struct __py
       __Pyx_GIVEREF(__pyx_t_5);
       __pyx_t_2 = 0;
       __pyx_t_5 = 0;
-      __pyx_t_5 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_113), ((PyObject *)__pyx_t_4)); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 450; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_5 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_112), ((PyObject *)__pyx_t_4)); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 450; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(((PyObject *)__pyx_t_5));
       __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
       __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 450; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
@@ -39811,9 +40008,9 @@ static PyObject *__pyx_pf_3_sa_23HieroCachingRuleFactory_8precompute(struct __py
       __Pyx_GOTREF(__pyx_t_5);
       __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 452; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_INCREF(((PyObject *)__pyx_kp_s_114));
-      PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_kp_s_114));
-      __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_114));
+      __Pyx_INCREF(((PyObject *)__pyx_kp_s_113));
+      PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_kp_s_113));
+      __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_113));
       PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_t_5);
       __Pyx_GIVEREF(__pyx_t_5);
       __pyx_t_5 = 0;
@@ -39965,9 +40162,9 @@ static PyObject *__pyx_pf_3_sa_23HieroCachingRuleFactory_8precompute(struct __py
       __Pyx_GOTREF(__pyx_t_5);
       __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 458; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_INCREF(((PyObject *)__pyx_kp_s_115));
-      PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_kp_s_115));
-      __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_115));
+      __Pyx_INCREF(((PyObject *)__pyx_kp_s_114));
+      PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_kp_s_114));
+      __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_114));
       PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_t_5);
       __Pyx_GIVEREF(__pyx_t_5);
       __pyx_t_5 = 0;
@@ -40071,9 +40268,9 @@ static PyObject *__pyx_pf_3_sa_23HieroCachingRuleFactory_8precompute(struct __py
     __Pyx_GOTREF(__pyx_t_5);
     __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 463; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_INCREF(((PyObject *)__pyx_kp_s_116));
-    PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_kp_s_116));
-    __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_116));
+    __Pyx_INCREF(((PyObject *)__pyx_kp_s_115));
+    PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_kp_s_115));
+    __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_115));
     PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_t_5);
     __Pyx_GIVEREF(__pyx_t_5);
     __pyx_t_5 = 0;
@@ -42793,8 +42990,8 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_loc2str(CYTHON_UNUSED st
  *         i = 0
  *         while i < loc.arr_high:
  */
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_117));
-  __pyx_v_result = ((PyObject *)__pyx_kp_s_117);
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_116));
+  __pyx_v_result = ((PyObject *)__pyx_kp_s_116);
 
   /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":811
  *         cdef int i, j
@@ -42823,7 +43020,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_loc2str(CYTHON_UNUSED st
  *             for j from i <= j < i + loc.num_subpatterns:
  *                 result = result + ("%d " %loc.arr[j])
  */
-    __pyx_t_2 = PyNumber_Add(__pyx_v_result, ((PyObject *)__pyx_kp_s_118)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 813; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyNumber_Add(__pyx_v_result, ((PyObject *)__pyx_kp_s_117)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 813; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_DECREF(__pyx_v_result);
     __pyx_v_result = __pyx_t_2;
@@ -42889,7 +43086,7 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_loc2str(CYTHON_UNUSED st
  *         return result
  * 
  */
-  __pyx_t_2 = PyNumber_Add(__pyx_v_result, ((PyObject *)__pyx_kp_s_119)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 818; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyNumber_Add(__pyx_v_result, ((PyObject *)__pyx_kp_s_118)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 818; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_v_result);
   __pyx_v_result = __pyx_t_2;
@@ -43006,7 +43203,7 @@ static struct __pyx_obj_3_sa_PhraseLocation *__pyx_f_3_sa_23HieroCachingRuleFact
  *         if result is not None:
  *             intersect_method = "precomputed"
  */
-  __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self), __pyx_n_s_120); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 830; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(((PyObject *)__pyx_v_self), __pyx_n_s_119); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 830; __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 = 830; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
@@ -43084,9 +43281,9 @@ static struct __pyx_obj_3_sa_PhraseLocation *__pyx_f_3_sa_23HieroCachingRuleFact
  *             else:
  *                 result = self.intersect_helper(prefix, suffix, prefix_loc, suffix_loc, MERGE)
  */
-      __Pyx_INCREF(((PyObject *)__pyx_kp_s_121));
+      __Pyx_INCREF(((PyObject *)__pyx_kp_s_120));
       __Pyx_XDECREF(__pyx_v_intersect_method);
-      __pyx_v_intersect_method = ((PyObject *)__pyx_kp_s_121);
+      __pyx_v_intersect_method = ((PyObject *)__pyx_kp_s_120);
       goto __pyx_L5;
     }
     /*else*/ {
@@ -46783,7 +46980,7 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
  *                 # checking whether lookup_required
  *                 if lookup_required:
  */
-            __pyx_t_3 = PyObject_Call(__pyx_builtin_Exception, ((PyObject *)__pyx_k_tuple_123), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1022; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_3 = PyObject_Call(__pyx_builtin_Exception, ((PyObject *)__pyx_k_tuple_122), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1022; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_3);
             __Pyx_Raise(__pyx_t_3, 0, 0, 0);
             __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
@@ -48599,7 +48796,7 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
  *                             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 = 1127; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_9 = PyObject_GetAttr(((PyObject *)__pyx_cur_scope->__pyx_v_self), __pyx_n_s_123); 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 = 1127; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
             __Pyx_GOTREF(__pyx_t_10);
@@ -48863,9 +49060,9 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
   __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 = 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));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_124));
+  PyTuple_SET_ITEM(__pyx_t_12, 0, ((PyObject *)__pyx_kp_s_124));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_124));
   PyTuple_SET_ITEM(__pyx_t_12, 1, __pyx_t_13);
   __Pyx_GIVEREF(__pyx_t_13);
   __pyx_t_13 = 0;
@@ -48908,9 +49105,9 @@ static PyObject *__pyx_gb_3_sa_23HieroCachingRuleFactory_24generator4(__pyx_Gene
   __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));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_125));
+  PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_kp_s_125));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_125));
   PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_t_13);
   __Pyx_GIVEREF(__pyx_t_13);
   __pyx_t_13 = 0;
@@ -52078,9 +52275,9 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  *                 met_constraints = 0
  *             if self.require_aligned_chunks and num_aligned_chunks < num_chunks:
  */
-      __Pyx_INCREF(((PyObject *)__pyx_kp_s_127));
+      __Pyx_INCREF(((PyObject *)__pyx_kp_s_126));
       __Pyx_DECREF(__pyx_v_reason_for_failure);
-      __pyx_v_reason_for_failure = ((PyObject *)__pyx_kp_s_127);
+      __pyx_v_reason_for_failure = ((PyObject *)__pyx_kp_s_126);
 
       /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1496
  *             if num_aligned_chunks == 0:
@@ -52116,9 +52313,9 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  *                 met_constraints = 0
  * 
  */
-      __Pyx_INCREF(((PyObject *)__pyx_kp_s_128));
+      __Pyx_INCREF(((PyObject *)__pyx_kp_s_127));
       __Pyx_DECREF(__pyx_v_reason_for_failure);
-      __pyx_v_reason_for_failure = ((PyObject *)__pyx_kp_s_128);
+      __pyx_v_reason_for_failure = ((PyObject *)__pyx_kp_s_127);
 
       /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1499
  *             if self.require_aligned_chunks and num_aligned_chunks < num_chunks:
@@ -52176,9 +52373,9 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  *                     met_constraints = 0
  *                     break
  */
-        __Pyx_INCREF(((PyObject *)__pyx_kp_s_129));
+        __Pyx_INCREF(((PyObject *)__pyx_kp_s_128));
         __Pyx_DECREF(__pyx_v_reason_for_failure);
-        __pyx_v_reason_for_failure = ((PyObject *)__pyx_kp_s_129);
+        __pyx_v_reason_for_failure = ((PyObject *)__pyx_kp_s_128);
 
         /* "/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:
@@ -52218,9 +52415,9 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  *                     met_constraints = 0
  *                     break
  */
-        __Pyx_INCREF(((PyObject *)__pyx_kp_s_129));
+        __Pyx_INCREF(((PyObject *)__pyx_kp_s_128));
         __Pyx_DECREF(__pyx_v_reason_for_failure);
-        __pyx_v_reason_for_failure = ((PyObject *)__pyx_kp_s_129);
+        __pyx_v_reason_for_failure = ((PyObject *)__pyx_kp_s_128);
 
         /* "/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:
@@ -52432,9 +52629,9 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  *                 else:
  *                     gap_start = 1
  */
-            __Pyx_INCREF(((PyObject *)__pyx_kp_s_130));
+            __Pyx_INCREF(((PyObject *)__pyx_kp_s_129));
             __Pyx_DECREF(__pyx_v_reason_for_failure);
-            __pyx_v_reason_for_failure = ((PyObject *)__pyx_kp_s_130);
+            __pyx_v_reason_for_failure = ((PyObject *)__pyx_kp_s_129);
             goto __pyx_L38;
           }
           __pyx_L38:;
@@ -52630,9 +52827,9 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  *                 else:
  *                     if self.tight_phrases and f_links_low[f_high-1] == -1:
  */
-            __Pyx_INCREF(((PyObject *)__pyx_kp_s_131));
+            __Pyx_INCREF(((PyObject *)__pyx_kp_s_130));
             __Pyx_DECREF(__pyx_v_reason_for_failure);
-            __pyx_v_reason_for_failure = ((PyObject *)__pyx_kp_s_131);
+            __pyx_v_reason_for_failure = ((PyObject *)__pyx_kp_s_130);
             goto __pyx_L45;
           }
           __pyx_L45:;
@@ -52750,7 +52947,7 @@ 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 = 1575; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_1 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_131), ((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);
@@ -53066,7 +53263,7 @@ 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 = 1605; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+            __pyx_t_14 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_132), ((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);
@@ -55284,9 +55481,9 @@ static PyObject *__pyx_f_3_sa_23HieroCachingRuleFactory_extract(struct __pyx_obj
  * 
  *         free(sent_links)
  */
-      __Pyx_INCREF(((PyObject *)__pyx_kp_s_134));
+      __Pyx_INCREF(((PyObject *)__pyx_kp_s_133));
       __Pyx_DECREF(__pyx_v_reason_for_failure);
-      __pyx_v_reason_for_failure = ((PyObject *)__pyx_kp_s_134);
+      __pyx_v_reason_for_failure = ((PyObject *)__pyx_kp_s_133);
     }
     __pyx_L34:;
     goto __pyx_L33;
@@ -55887,7 +56084,7 @@ static PyObject *__pyx_gb_3_sa_13FeatureVector_7__str___2generator12(__pyx_Gener
     __Pyx_GIVEREF(__pyx_t_4);
     __pyx_cur_scope->__pyx_v_feat = __pyx_t_4;
     __pyx_t_4 = 0;
-    __pyx_t_4 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_135), __pyx_cur_scope->__pyx_v_feat); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[13]; __pyx_lineno = 21; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_134), __pyx_cur_scope->__pyx_v_feat); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[13]; __pyx_lineno = 21; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_4));
     __pyx_r = ((PyObject *)__pyx_t_4);
     __pyx_t_4 = 0;
@@ -57617,6 +57814,39 @@ static int __pyx_tp_clear_3_sa_DataArray(PyObject *o) {
   Py_XDECREF(tmp);
   return 0;
 }
+static PyObject *__pyx_sq_item_3_sa_DataArray(PyObject *o, Py_ssize_t i) {
+  PyObject *r;
+  PyObject *x = PyInt_FromSsize_t(i); if(!x) return 0;
+  r = Py_TYPE(o)->tp_as_mapping->mp_subscript(o, x);
+  Py_DECREF(x);
+  return r;
+}
+
+static PyObject *__pyx_getprop_3_sa_9DataArray_word2id(PyObject *o, CYTHON_UNUSED void *x) {
+  return __pyx_pw_3_sa_9DataArray_7word2id_1__get__(o);
+}
+
+static int __pyx_setprop_3_sa_9DataArray_word2id(PyObject *o, PyObject *v, CYTHON_UNUSED void *x) {
+  if (v) {
+    return __pyx_pw_3_sa_9DataArray_7word2id_3__set__(o, v);
+  }
+  else {
+    return __pyx_pw_3_sa_9DataArray_7word2id_5__del__(o);
+  }
+}
+
+static PyObject *__pyx_getprop_3_sa_9DataArray_id2word(PyObject *o, CYTHON_UNUSED void *x) {
+  return __pyx_pw_3_sa_9DataArray_7id2word_1__get__(o);
+}
+
+static int __pyx_setprop_3_sa_9DataArray_id2word(PyObject *o, PyObject *v, CYTHON_UNUSED void *x) {
+  if (v) {
+    return __pyx_pw_3_sa_9DataArray_7id2word_3__set__(o, v);
+  }
+  else {
+    return __pyx_pw_3_sa_9DataArray_7id2word_5__del__(o);
+  }
+}
 
 static PyObject *__pyx_getprop_3_sa_9DataArray_data(PyObject *o, CYTHON_UNUSED void *x) {
   return __pyx_pw_3_sa_9DataArray_4data_1__get__(o);
@@ -57631,12 +57861,37 @@ static int __pyx_setprop_3_sa_9DataArray_data(PyObject *o, PyObject *v, CYTHON_U
   }
 }
 
+static PyObject *__pyx_getprop_3_sa_9DataArray_sent_id(PyObject *o, CYTHON_UNUSED void *x) {
+  return __pyx_pw_3_sa_9DataArray_7sent_id_1__get__(o);
+}
+
+static int __pyx_setprop_3_sa_9DataArray_sent_id(PyObject *o, PyObject *v, CYTHON_UNUSED void *x) {
+  if (v) {
+    return __pyx_pw_3_sa_9DataArray_7sent_id_3__set__(o, v);
+  }
+  else {
+    return __pyx_pw_3_sa_9DataArray_7sent_id_5__del__(o);
+  }
+}
+
+static PyObject *__pyx_getprop_3_sa_9DataArray_sent_index(PyObject *o, CYTHON_UNUSED void *x) {
+  return __pyx_pw_3_sa_9DataArray_10sent_index_1__get__(o);
+}
+
+static int __pyx_setprop_3_sa_9DataArray_sent_index(PyObject *o, PyObject *v, CYTHON_UNUSED void *x) {
+  if (v) {
+    return __pyx_pw_3_sa_9DataArray_10sent_index_3__set__(o, v);
+  }
+  else {
+    return __pyx_pw_3_sa_9DataArray_10sent_index_5__del__(o);
+  }
+}
+
 static PyMethodDef __pyx_methods_3_sa_DataArray[] = {
   {__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("get_id"), (PyCFunction)__pyx_pw_3_sa_9DataArray_9get_id, METH_O, __Pyx_DOCSTR(0)},
+  {__Pyx_NAMESTR("get_sentence_bounds"), (PyCFunction)__pyx_pw_3_sa_9DataArray_13get_sentence_bounds, 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)},
@@ -57649,7 +57904,11 @@ static PyMethodDef __pyx_methods_3_sa_DataArray[] = {
 };
 
 static struct PyGetSetDef __pyx_getsets_3_sa_DataArray[] = {
+  {(char *)"word2id", __pyx_getprop_3_sa_9DataArray_word2id, __pyx_setprop_3_sa_9DataArray_word2id, 0, 0},
+  {(char *)"id2word", __pyx_getprop_3_sa_9DataArray_id2word, __pyx_setprop_3_sa_9DataArray_id2word, 0, 0},
   {(char *)"data", __pyx_getprop_3_sa_9DataArray_data, __pyx_setprop_3_sa_9DataArray_data, 0, 0},
+  {(char *)"sent_id", __pyx_getprop_3_sa_9DataArray_sent_id, __pyx_setprop_3_sa_9DataArray_sent_id, 0, 0},
+  {(char *)"sent_index", __pyx_getprop_3_sa_9DataArray_sent_index, __pyx_setprop_3_sa_9DataArray_sent_index, 0, 0},
   {0, 0, 0, 0, 0}
 };
 
@@ -57715,7 +57974,7 @@ static PySequenceMethods __pyx_tp_as_sequence_DataArray = {
   __pyx_pw_3_sa_9DataArray_3__len__, /*sq_length*/
   0, /*sq_concat*/
   0, /*sq_repeat*/
-  0, /*sq_item*/
+  __pyx_sq_item_3_sa_DataArray, /*sq_item*/
   0, /*sq_slice*/
   0, /*sq_ass_item*/
   0, /*sq_ass_slice*/
@@ -57726,7 +57985,7 @@ static PySequenceMethods __pyx_tp_as_sequence_DataArray = {
 
 static PyMappingMethods __pyx_tp_as_mapping_DataArray = {
   __pyx_pw_3_sa_9DataArray_3__len__, /*mp_length*/
-  0, /*mp_subscript*/
+  __pyx_pw_3_sa_9DataArray_11__getitem__, /*mp_subscript*/
   0, /*mp_ass_subscript*/
 };
 
@@ -59881,16 +60140,13 @@ static PyObject *__pyx_sq_item_3_sa_SuffixArray(PyObject *o, Py_ssize_t i) {
 }
 
 static PyMethodDef __pyx_methods_3_sa_SuffixArray[] = {
-  {__Pyx_NAMESTR("get_sentence_id"), (PyCFunction)__pyx_pw_3_sa_11SuffixArray_5get_sentence_id, METH_O, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("get_sentence"), (PyCFunction)__pyx_pw_3_sa_11SuffixArray_7get_sentence, METH_O, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("get_sentence_position"), (PyCFunction)__pyx_pw_3_sa_11SuffixArray_9get_sentence_position, METH_O, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("read_text"), (PyCFunction)__pyx_pw_3_sa_11SuffixArray_11read_text, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_3_sa_11SuffixArray_10read_text)},
-  {__Pyx_NAMESTR("q3sort"), (PyCFunction)__pyx_pw_3_sa_11SuffixArray_13q3sort, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_3_sa_11SuffixArray_12q3sort)},
-  {__Pyx_NAMESTR("write_text"), (PyCFunction)__pyx_pw_3_sa_11SuffixArray_15write_text, METH_O, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("read_binary"), (PyCFunction)__pyx_pw_3_sa_11SuffixArray_17read_binary, METH_O, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("write_binary"), (PyCFunction)__pyx_pw_3_sa_11SuffixArray_19write_binary, METH_O, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("write_enhanced"), (PyCFunction)__pyx_pw_3_sa_11SuffixArray_21write_enhanced, METH_O, __Pyx_DOCSTR(0)},
-  {__Pyx_NAMESTR("lookup"), (PyCFunction)__pyx_pw_3_sa_11SuffixArray_23lookup, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(0)},
+  {__Pyx_NAMESTR("read_text"), (PyCFunction)__pyx_pw_3_sa_11SuffixArray_5read_text, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_3_sa_11SuffixArray_4read_text)},
+  {__Pyx_NAMESTR("q3sort"), (PyCFunction)__pyx_pw_3_sa_11SuffixArray_7q3sort, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_3_sa_11SuffixArray_6q3sort)},
+  {__Pyx_NAMESTR("write_text"), (PyCFunction)__pyx_pw_3_sa_11SuffixArray_9write_text, METH_O, __Pyx_DOCSTR(0)},
+  {__Pyx_NAMESTR("read_binary"), (PyCFunction)__pyx_pw_3_sa_11SuffixArray_11read_binary, METH_O, __Pyx_DOCSTR(0)},
+  {__Pyx_NAMESTR("write_binary"), (PyCFunction)__pyx_pw_3_sa_11SuffixArray_13write_binary, METH_O, __Pyx_DOCSTR(0)},
+  {__Pyx_NAMESTR("write_enhanced"), (PyCFunction)__pyx_pw_3_sa_11SuffixArray_15write_enhanced, METH_O, __Pyx_DOCSTR(0)},
+  {__Pyx_NAMESTR("lookup"), (PyCFunction)__pyx_pw_3_sa_11SuffixArray_17lookup, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(0)},
   {0, 0, 0, 0}
 };
 
@@ -65928,11 +66184,12 @@ static struct PyModuleDef __pyx_moduledef = {
 
 static __Pyx_StringTabEntry __pyx_string_tab[] = {
   {&__pyx_kp_s_1, __pyx_k_1, sizeof(__pyx_k_1), 0, 0, 1, 0},
+  {&__pyx_kp_s_100, __pyx_k_100, sizeof(__pyx_k_100), 0, 0, 1, 0},
   {&__pyx_kp_s_101, __pyx_k_101, sizeof(__pyx_k_101), 0, 0, 1, 0},
-  {&__pyx_kp_s_102, __pyx_k_102, sizeof(__pyx_k_102), 0, 0, 1, 0},
+  {&__pyx_n_s_103, __pyx_k_103, sizeof(__pyx_k_103), 0, 0, 1, 1},
   {&__pyx_n_s_104, __pyx_k_104, sizeof(__pyx_k_104), 0, 0, 1, 1},
-  {&__pyx_n_s_105, __pyx_k_105, sizeof(__pyx_k_105), 0, 0, 1, 1},
-  {&__pyx_kp_s_107, __pyx_k_107, sizeof(__pyx_k_107), 0, 0, 1, 0},
+  {&__pyx_kp_s_106, __pyx_k_106, sizeof(__pyx_k_106), 0, 0, 1, 0},
+  {&__pyx_kp_s_108, __pyx_k_108, sizeof(__pyx_k_108), 0, 0, 1, 0},
   {&__pyx_kp_s_109, __pyx_k_109, sizeof(__pyx_k_109), 0, 0, 1, 0},
   {&__pyx_kp_s_110, __pyx_k_110, sizeof(__pyx_k_110), 0, 0, 1, 0},
   {&__pyx_kp_s_111, __pyx_k_111, sizeof(__pyx_k_111), 0, 0, 1, 0},
@@ -65943,11 +66200,11 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = {
   {&__pyx_kp_s_116, __pyx_k_116, sizeof(__pyx_k_116), 0, 0, 1, 0},
   {&__pyx_kp_s_117, __pyx_k_117, sizeof(__pyx_k_117), 0, 0, 1, 0},
   {&__pyx_kp_s_118, __pyx_k_118, sizeof(__pyx_k_118), 0, 0, 1, 0},
-  {&__pyx_kp_s_119, __pyx_k_119, sizeof(__pyx_k_119), 0, 0, 1, 0},
-  {&__pyx_n_s_120, __pyx_k_120, sizeof(__pyx_k_120), 0, 0, 1, 1},
+  {&__pyx_n_s_119, __pyx_k_119, sizeof(__pyx_k_119), 0, 0, 1, 1},
+  {&__pyx_kp_s_120, __pyx_k_120, sizeof(__pyx_k_120), 0, 0, 1, 0},
   {&__pyx_kp_s_121, __pyx_k_121, sizeof(__pyx_k_121), 0, 0, 1, 0},
-  {&__pyx_kp_s_122, __pyx_k_122, sizeof(__pyx_k_122), 0, 0, 1, 0},
-  {&__pyx_n_s_124, __pyx_k_124, sizeof(__pyx_k_124), 0, 0, 1, 1},
+  {&__pyx_n_s_123, __pyx_k_123, sizeof(__pyx_k_123), 0, 0, 1, 1},
+  {&__pyx_kp_s_124, __pyx_k_124, sizeof(__pyx_k_124), 0, 0, 1, 0},
   {&__pyx_kp_s_125, __pyx_k_125, sizeof(__pyx_k_125), 0, 0, 1, 0},
   {&__pyx_kp_s_126, __pyx_k_126, sizeof(__pyx_k_126), 0, 0, 1, 0},
   {&__pyx_kp_s_127, __pyx_k_127, sizeof(__pyx_k_127), 0, 0, 1, 0},
@@ -65959,11 +66216,10 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = {
   {&__pyx_kp_s_132, __pyx_k_132, sizeof(__pyx_k_132), 0, 0, 1, 0},
   {&__pyx_kp_s_133, __pyx_k_133, sizeof(__pyx_k_133), 0, 0, 1, 0},
   {&__pyx_kp_s_134, __pyx_k_134, sizeof(__pyx_k_134), 0, 0, 1, 0},
-  {&__pyx_kp_s_135, __pyx_k_135, sizeof(__pyx_k_135), 0, 0, 1, 0},
+  {&__pyx_kp_s_137, __pyx_k_137, sizeof(__pyx_k_137), 0, 0, 1, 0},
   {&__pyx_kp_s_138, __pyx_k_138, sizeof(__pyx_k_138), 0, 0, 1, 0},
-  {&__pyx_kp_s_139, __pyx_k_139, sizeof(__pyx_k_139), 0, 0, 1, 0},
   {&__pyx_kp_s_14, __pyx_k_14, sizeof(__pyx_k_14), 0, 0, 1, 0},
-  {&__pyx_kp_s_143, __pyx_k_143, sizeof(__pyx_k_143), 0, 0, 1, 0},
+  {&__pyx_kp_s_142, __pyx_k_142, sizeof(__pyx_k_142), 0, 0, 1, 0},
   {&__pyx_kp_s_18, __pyx_k_18, sizeof(__pyx_k_18), 0, 0, 1, 0},
   {&__pyx_kp_s_2, __pyx_k_2, sizeof(__pyx_k_2), 0, 0, 1, 0},
   {&__pyx_kp_s_21, __pyx_k_21, sizeof(__pyx_k_21), 0, 0, 1, 0},
@@ -66006,14 +66262,13 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = {
   {&__pyx_kp_s_86, __pyx_k_86, sizeof(__pyx_k_86), 0, 0, 1, 0},
   {&__pyx_kp_s_87, __pyx_k_87, sizeof(__pyx_k_87), 0, 0, 1, 0},
   {&__pyx_kp_s_88, __pyx_k_88, sizeof(__pyx_k_88), 0, 0, 1, 0},
-  {&__pyx_n_s_89, __pyx_k_89, sizeof(__pyx_k_89), 0, 0, 1, 1},
+  {&__pyx_kp_s_89, __pyx_k_89, sizeof(__pyx_k_89), 0, 0, 1, 0},
   {&__pyx_kp_s_9, __pyx_k_9, sizeof(__pyx_k_9), 0, 0, 1, 0},
   {&__pyx_kp_s_90, __pyx_k_90, sizeof(__pyx_k_90), 0, 0, 1, 0},
   {&__pyx_kp_s_91, __pyx_k_91, sizeof(__pyx_k_91), 0, 0, 1, 0},
   {&__pyx_kp_s_92, __pyx_k_92, sizeof(__pyx_k_92), 0, 0, 1, 0},
-  {&__pyx_kp_s_93, __pyx_k_93, sizeof(__pyx_k_93), 0, 0, 1, 0},
+  {&__pyx_kp_s_94, __pyx_k_94, sizeof(__pyx_k_94), 0, 0, 1, 0},
   {&__pyx_kp_s_95, __pyx_k_95, sizeof(__pyx_k_95), 0, 0, 1, 0},
-  {&__pyx_kp_s_96, __pyx_k_96, sizeof(__pyx_k_96), 0, 0, 1, 0},
   {&__pyx_kp_s__0, __pyx_k__0, sizeof(__pyx_k__0), 0, 0, 1, 0},
   {&__pyx_kp_s__1, __pyx_k__1, sizeof(__pyx_k__1), 0, 0, 1, 0},
   {&__pyx_n_s__Counter, __pyx_k__Counter, sizeof(__pyx_k__Counter), 0, 0, 1, 1},
@@ -66091,8 +66346,6 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = {
   {&__pyx_n_s__get_f_id, __pyx_k__get_f_id, sizeof(__pyx_k__get_f_id), 0, 0, 1, 1},
   {&__pyx_n_s__get_id, __pyx_k__get_id, sizeof(__pyx_k__get_id), 0, 0, 1, 1},
   {&__pyx_n_s__get_next_states, __pyx_k__get_next_states, sizeof(__pyx_k__get_next_states), 0, 0, 1, 1},
-  {&__pyx_n_s__get_sentence, __pyx_k__get_sentence, sizeof(__pyx_k__get_sentence), 0, 0, 1, 1},
-  {&__pyx_n_s__get_sentence_id, __pyx_k__get_sentence_id, sizeof(__pyx_k__get_sentence_id), 0, 0, 1, 1},
   {&__pyx_n_s__get_word, __pyx_k__get_word, sizeof(__pyx_k__get_word), 0, 0, 1, 1},
   {&__pyx_n_s__getchunk, __pyx_k__getchunk, sizeof(__pyx_k__getchunk), 0, 0, 1, 1},
   {&__pyx_n_s__getrusage, __pyx_k__getrusage, sizeof(__pyx_k__getrusage), 0, 0, 1, 1},
@@ -66215,7 +66468,7 @@ 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 = 78; __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 = 79; __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;}
@@ -66283,28 +66536,28 @@ 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":65
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":66
  *                     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 = 65; __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 = 66; __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":60
+  /* "/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:
  *                 if w_id > 1:
  */
-  __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_k_tuple_16 = PyTuple_New(3); if (unlikely(!__pyx_k_tuple_16)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 61; __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);
@@ -66317,14 +66570,14 @@ 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":68
+  /* "/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)
  * 
  */
-  __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_k_tuple_17 = PyTuple_New(3); if (unlikely(!__pyx_k_tuple_17)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 69; __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);
@@ -66337,28 +66590,28 @@ 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":73
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":74
  *     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 = 73; __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 = 74; __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":72
+  /* "/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_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_k_tuple_20 = PyTuple_New(3); if (unlikely(!__pyx_k_tuple_20)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 73; __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);
@@ -66371,69 +66624,69 @@ 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":143
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":144
  *         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 = 143; __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 = 144; __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":146
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":147
  *         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 = 146; __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 = 147; __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":149
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":150
  *         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 = 149; __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 = 150; __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":152
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/data_array.pxi":153
  *         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 = 152; __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 = 153; __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":155
+  /* "/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_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_k_tuple_28 = PyTuple_New(3); if (unlikely(!__pyx_k_tuple_28)) {__pyx_filename = __pyx_f[3]; __pyx_lineno = 156; __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);
@@ -66880,67 +67133,67 @@ static int __Pyx_InitCachedConstants(void) {
   __Pyx_GIVEREF(__pyx_int_neg_1);
   __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_82));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":103
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":94
  * 
  *         '''Step 3: read off suffix array from inverse suffix array'''
  *         logger.info("    Finalizing sort...")             # <<<<<<<<<<<<<<
  *         for i from 0 <= i < N:
  *             j = isa.arr[i]
  */
-  __pyx_k_tuple_94 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_94)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 103; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_k_tuple_94);
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_93));
-  PyTuple_SET_ITEM(__pyx_k_tuple_94, 0, ((PyObject *)__pyx_kp_s_93));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_93));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_94));
+  __pyx_k_tuple_93 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_93)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 94; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_k_tuple_93);
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_92));
+  PyTuple_SET_ITEM(__pyx_k_tuple_93, 0, ((PyObject *)__pyx_kp_s_92));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_92));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_93));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":202
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":193
  *             for a_i in self.sa:
  *                 f.write("%d " % a_i)
  *             f.write("\n")             # <<<<<<<<<<<<<<
  *             for w_i in self.ha:
  *                 f.write("%d " % w_i)
  */
-  __pyx_k_tuple_97 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_97)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 202; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_k_tuple_97);
+  __pyx_k_tuple_96 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_96)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 193; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_k_tuple_96);
   __Pyx_INCREF(((PyObject *)__pyx_kp_s_14));
-  PyTuple_SET_ITEM(__pyx_k_tuple_97, 0, ((PyObject *)__pyx_kp_s_14));
+  PyTuple_SET_ITEM(__pyx_k_tuple_96, 0, ((PyObject *)__pyx_kp_s_14));
   __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_14));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_97));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_96));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":205
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":196
  *             for w_i in self.ha:
  *                 f.write("%d " % w_i)
  *             f.write("\n")             # <<<<<<<<<<<<<<
  * 
  *     cdef int __search_high(self, int word_id, int offset, int low, int high):
  */
-  __pyx_k_tuple_98 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_98)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 205; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_k_tuple_98);
+  __pyx_k_tuple_97 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_97)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 196; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_k_tuple_97);
   __Pyx_INCREF(((PyObject *)__pyx_kp_s_14));
-  PyTuple_SET_ITEM(__pyx_k_tuple_98, 0, ((PyObject *)__pyx_kp_s_14));
+  PyTuple_SET_ITEM(__pyx_k_tuple_97, 0, ((PyObject *)__pyx_kp_s_14));
   __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_14));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_98));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_97));
 
-  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":198
+  /* "/Users/vchahun/Sandbox/cdec/python/src/sa/suffix_array.pxi":189
  * 
  *     def write_enhanced(self, char* filename):
  *         with open(filename, "w") as f:             # <<<<<<<<<<<<<<
  *             self.darray.write_enhanced_handle(f)
  *             for a_i in self.sa:
  */
-  __pyx_k_tuple_99 = PyTuple_New(3); if (unlikely(!__pyx_k_tuple_99)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 198; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_k_tuple_99);
+  __pyx_k_tuple_98 = PyTuple_New(3); if (unlikely(!__pyx_k_tuple_98)) {__pyx_filename = __pyx_f[12]; __pyx_lineno = 189; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_k_tuple_98);
   __Pyx_INCREF(Py_None);
-  PyTuple_SET_ITEM(__pyx_k_tuple_99, 0, Py_None);
+  PyTuple_SET_ITEM(__pyx_k_tuple_98, 0, Py_None);
   __Pyx_GIVEREF(Py_None);
   __Pyx_INCREF(Py_None);
-  PyTuple_SET_ITEM(__pyx_k_tuple_99, 1, Py_None);
+  PyTuple_SET_ITEM(__pyx_k_tuple_98, 1, Py_None);
   __Pyx_GIVEREF(Py_None);
   __Pyx_INCREF(Py_None);
-  PyTuple_SET_ITEM(__pyx_k_tuple_99, 2, Py_None);
+  PyTuple_SET_ITEM(__pyx_k_tuple_98, 2, Py_None);
   __Pyx_GIVEREF(Py_None);
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_99));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_98));
 
   /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":107
  *             logger.info("Sampling strategy: uniform, max sample size = %d", sample_size)
@@ -66949,12 +67202,12 @@ static int __Pyx_InitCachedConstants(void) {
  * 
  *     def sample(self, PhraseLocation phrase_location):
  */
-  __pyx_k_tuple_103 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_103)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_k_tuple_103);
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_102));
-  PyTuple_SET_ITEM(__pyx_k_tuple_103, 0, ((PyObject *)__pyx_kp_s_102));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_102));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_103));
+  __pyx_k_tuple_102 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_102)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_k_tuple_102);
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_101));
+  PyTuple_SET_ITEM(__pyx_k_tuple_102, 0, ((PyObject *)__pyx_kp_s_101));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_101));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_102));
 
   /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":316
  *         self.rules.root = ExtendedTrieNode(phrase_location=PhraseLocation())
@@ -66963,12 +67216,12 @@ static int __Pyx_InitCachedConstants(void) {
  *         self.alignment = alignment
  * 
  */
-  __pyx_k_tuple_108 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_108)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 316; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_k_tuple_108);
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_107));
-  PyTuple_SET_ITEM(__pyx_k_tuple_108, 0, ((PyObject *)__pyx_kp_s_107));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_107));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_108));
+  __pyx_k_tuple_107 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_107)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 316; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_k_tuple_107);
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_106));
+  PyTuple_SET_ITEM(__pyx_k_tuple_107, 0, ((PyObject *)__pyx_kp_s_106));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_106));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_107));
 
   /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":1022
  *                         else:
@@ -66977,12 +67230,12 @@ static int __Pyx_InitCachedConstants(void) {
  *                 # checking whether lookup_required
  *                 if lookup_required:
  */
-  __pyx_k_tuple_123 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_123)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1022; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_k_tuple_123);
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_122));
-  PyTuple_SET_ITEM(__pyx_k_tuple_123, 0, ((PyObject *)__pyx_kp_s_122));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_122));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_123));
+  __pyx_k_tuple_122 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_122)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 1022; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_k_tuple_122);
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_121));
+  PyTuple_SET_ITEM(__pyx_k_tuple_122, 0, ((PyObject *)__pyx_kp_s_121));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_121));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_122));
 
   /* "_sa.pyx":9
  *             resource.getrusage(resource.RUSAGE_SELF).ru_stime)
@@ -66991,16 +67244,16 @@ static int __Pyx_InitCachedConstants(void) {
  *     if filename.endswith('.gz'):
  *         return gzip.GzipFile(filename)
  */
-  __pyx_k_tuple_136 = PyTuple_New(2); if (unlikely(!__pyx_k_tuple_136)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 9; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_k_tuple_136);
+  __pyx_k_tuple_135 = PyTuple_New(2); if (unlikely(!__pyx_k_tuple_135)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 9; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_k_tuple_135);
   __Pyx_INCREF(((PyObject *)__pyx_n_s__filename));
-  PyTuple_SET_ITEM(__pyx_k_tuple_136, 0, ((PyObject *)__pyx_n_s__filename));
+  PyTuple_SET_ITEM(__pyx_k_tuple_135, 0, ((PyObject *)__pyx_n_s__filename));
   __Pyx_GIVEREF(((PyObject *)__pyx_n_s__filename));
   __Pyx_INCREF(((PyObject *)__pyx_n_s__filename));
-  PyTuple_SET_ITEM(__pyx_k_tuple_136, 1, ((PyObject *)__pyx_n_s__filename));
+  PyTuple_SET_ITEM(__pyx_k_tuple_135, 1, ((PyObject *)__pyx_n_s__filename));
   __Pyx_GIVEREF(((PyObject *)__pyx_n_s__filename));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_136));
-  __pyx_k_codeobj_137 = (PyObject*)__Pyx_PyCode_New(1, 0, 2, 0, 0, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_k_tuple_136, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_138, __pyx_n_s__gzip_or_text, 9, __pyx_empty_bytes); if (unlikely(!__pyx_k_codeobj_137)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 9; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_135));
+  __pyx_k_codeobj_136 = (PyObject*)__Pyx_PyCode_New(1, 0, 2, 0, 0, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_k_tuple_135, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_137, __pyx_n_s__gzip_or_text, 9, __pyx_empty_bytes); if (unlikely(!__pyx_k_codeobj_136)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 9; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
 
   /* "_sa.pyx":15
  *         return open(filename)
@@ -67009,12 +67262,12 @@ static int __Pyx_InitCachedConstants(void) {
  * 
  * include "float_list.pxi"
  */
-  __pyx_k_tuple_140 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_140)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 15; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_k_tuple_140);
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_139));
-  PyTuple_SET_ITEM(__pyx_k_tuple_140, 0, ((PyObject *)__pyx_kp_s_139));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_139));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_140));
+  __pyx_k_tuple_139 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_139)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 15; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_k_tuple_139);
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_138));
+  PyTuple_SET_ITEM(__pyx_k_tuple_139, 0, ((PyObject *)__pyx_kp_s_138));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_138));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_139));
 
   /* "/Users/vchahun/Sandbox/cdec/python/src/sa/sym.pxi":107
  *     return ALPHABET.fromstring(string, terminal)
@@ -67023,25 +67276,25 @@ static int __Pyx_InitCachedConstants(void) {
  *     word_ids = (sym_fromstring(word, True) for word in words)
  *     return tuple(((word, None, 1), ) for word in word_ids)
  */
-  __pyx_k_tuple_141 = PyTuple_New(5); if (unlikely(!__pyx_k_tuple_141)) {__pyx_filename = __pyx_f[10]; __pyx_lineno = 107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_k_tuple_141);
+  __pyx_k_tuple_140 = PyTuple_New(5); if (unlikely(!__pyx_k_tuple_140)) {__pyx_filename = __pyx_f[10]; __pyx_lineno = 107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_k_tuple_140);
   __Pyx_INCREF(((PyObject *)__pyx_n_s__words));
-  PyTuple_SET_ITEM(__pyx_k_tuple_141, 0, ((PyObject *)__pyx_n_s__words));
+  PyTuple_SET_ITEM(__pyx_k_tuple_140, 0, ((PyObject *)__pyx_n_s__words));
   __Pyx_GIVEREF(((PyObject *)__pyx_n_s__words));
   __Pyx_INCREF(((PyObject *)__pyx_n_s__word_ids));
-  PyTuple_SET_ITEM(__pyx_k_tuple_141, 1, ((PyObject *)__pyx_n_s__word_ids));
+  PyTuple_SET_ITEM(__pyx_k_tuple_140, 1, ((PyObject *)__pyx_n_s__word_ids));
   __Pyx_GIVEREF(((PyObject *)__pyx_n_s__word_ids));
   __Pyx_INCREF(((PyObject *)__pyx_n_s__genexpr));
-  PyTuple_SET_ITEM(__pyx_k_tuple_141, 2, ((PyObject *)__pyx_n_s__genexpr));
+  PyTuple_SET_ITEM(__pyx_k_tuple_140, 2, ((PyObject *)__pyx_n_s__genexpr));
   __Pyx_GIVEREF(((PyObject *)__pyx_n_s__genexpr));
   __Pyx_INCREF(((PyObject *)__pyx_n_s__genexpr));
-  PyTuple_SET_ITEM(__pyx_k_tuple_141, 3, ((PyObject *)__pyx_n_s__genexpr));
+  PyTuple_SET_ITEM(__pyx_k_tuple_140, 3, ((PyObject *)__pyx_n_s__genexpr));
   __Pyx_GIVEREF(((PyObject *)__pyx_n_s__genexpr));
   __Pyx_INCREF(((PyObject *)__pyx_n_s__genexpr));
-  PyTuple_SET_ITEM(__pyx_k_tuple_141, 4, ((PyObject *)__pyx_n_s__genexpr));
+  PyTuple_SET_ITEM(__pyx_k_tuple_140, 4, ((PyObject *)__pyx_n_s__genexpr));
   __Pyx_GIVEREF(((PyObject *)__pyx_n_s__genexpr));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_141));
-  __pyx_k_codeobj_142 = (PyObject*)__Pyx_PyCode_New(1, 0, 5, 0, 0, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_k_tuple_141, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_143, __pyx_n_s__make_lattice, 107, __pyx_empty_bytes); if (unlikely(!__pyx_k_codeobj_142)) {__pyx_filename = __pyx_f[10]; __pyx_lineno = 107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_140));
+  __pyx_k_codeobj_141 = (PyObject*)__Pyx_PyCode_New(1, 0, 5, 0, 0, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_k_tuple_140, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_142, __pyx_n_s__make_lattice, 107, __pyx_empty_bytes); if (unlikely(!__pyx_k_codeobj_141)) {__pyx_filename = __pyx_f[10]; __pyx_lineno = 107; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
 
   /* "/Users/vchahun/Sandbox/cdec/python/src/sa/sym.pxi":111
  *     return tuple(((word, None, 1), ) for word in word_ids)
@@ -67050,19 +67303,19 @@ static int __Pyx_InitCachedConstants(void) {
  *     return tuple((sym_tostring(sym), weight, dist) for (sym, weight, dist) in arc
  *             for arc in node for node in lattice)
  */
-  __pyx_k_tuple_144 = PyTuple_New(3); if (unlikely(!__pyx_k_tuple_144)) {__pyx_filename = __pyx_f[10]; __pyx_lineno = 111; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_k_tuple_144);
+  __pyx_k_tuple_143 = PyTuple_New(3); if (unlikely(!__pyx_k_tuple_143)) {__pyx_filename = __pyx_f[10]; __pyx_lineno = 111; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_k_tuple_143);
   __Pyx_INCREF(((PyObject *)__pyx_n_s__lattice));
-  PyTuple_SET_ITEM(__pyx_k_tuple_144, 0, ((PyObject *)__pyx_n_s__lattice));
+  PyTuple_SET_ITEM(__pyx_k_tuple_143, 0, ((PyObject *)__pyx_n_s__lattice));
   __Pyx_GIVEREF(((PyObject *)__pyx_n_s__lattice));
   __Pyx_INCREF(((PyObject *)__pyx_n_s__genexpr));
-  PyTuple_SET_ITEM(__pyx_k_tuple_144, 1, ((PyObject *)__pyx_n_s__genexpr));
+  PyTuple_SET_ITEM(__pyx_k_tuple_143, 1, ((PyObject *)__pyx_n_s__genexpr));
   __Pyx_GIVEREF(((PyObject *)__pyx_n_s__genexpr));
   __Pyx_INCREF(((PyObject *)__pyx_n_s__genexpr));
-  PyTuple_SET_ITEM(__pyx_k_tuple_144, 2, ((PyObject *)__pyx_n_s__genexpr));
+  PyTuple_SET_ITEM(__pyx_k_tuple_143, 2, ((PyObject *)__pyx_n_s__genexpr));
   __Pyx_GIVEREF(((PyObject *)__pyx_n_s__genexpr));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_144));
-  __pyx_k_codeobj_145 = (PyObject*)__Pyx_PyCode_New(1, 0, 3, 0, 0, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_k_tuple_144, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_143, __pyx_n_s__decode_lattice, 111, __pyx_empty_bytes); if (unlikely(!__pyx_k_codeobj_145)) {__pyx_filename = __pyx_f[10]; __pyx_lineno = 111; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_143));
+  __pyx_k_codeobj_144 = (PyObject*)__Pyx_PyCode_New(1, 0, 3, 0, 0, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_k_tuple_143, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_142, __pyx_n_s__decode_lattice, 111, __pyx_empty_bytes); if (unlikely(!__pyx_k_codeobj_144)) {__pyx_filename = __pyx_f[10]; __pyx_lineno = 111; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
 
   /* "/Users/vchahun/Sandbox/cdec/python/src/sa/sym.pxi":115
  *             for arc in node for node in lattice)
@@ -67070,19 +67323,19 @@ static int __Pyx_InitCachedConstants(void) {
  * def decode_sentence(lattice):             # <<<<<<<<<<<<<<
  *     return tuple(sym_tostring(sym) for ((sym, _, _),) in lattice)
  */
-  __pyx_k_tuple_146 = PyTuple_New(3); if (unlikely(!__pyx_k_tuple_146)) {__pyx_filename = __pyx_f[10]; __pyx_lineno = 115; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_k_tuple_146);
+  __pyx_k_tuple_145 = PyTuple_New(3); if (unlikely(!__pyx_k_tuple_145)) {__pyx_filename = __pyx_f[10]; __pyx_lineno = 115; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_k_tuple_145);
   __Pyx_INCREF(((PyObject *)__pyx_n_s__lattice));
-  PyTuple_SET_ITEM(__pyx_k_tuple_146, 0, ((PyObject *)__pyx_n_s__lattice));
+  PyTuple_SET_ITEM(__pyx_k_tuple_145, 0, ((PyObject *)__pyx_n_s__lattice));
   __Pyx_GIVEREF(((PyObject *)__pyx_n_s__lattice));
   __Pyx_INCREF(((PyObject *)__pyx_n_s__genexpr));
-  PyTuple_SET_ITEM(__pyx_k_tuple_146, 1, ((PyObject *)__pyx_n_s__genexpr));
+  PyTuple_SET_ITEM(__pyx_k_tuple_145, 1, ((PyObject *)__pyx_n_s__genexpr));
   __Pyx_GIVEREF(((PyObject *)__pyx_n_s__genexpr));
   __Pyx_INCREF(((PyObject *)__pyx_n_s__genexpr));
-  PyTuple_SET_ITEM(__pyx_k_tuple_146, 2, ((PyObject *)__pyx_n_s__genexpr));
+  PyTuple_SET_ITEM(__pyx_k_tuple_145, 2, ((PyObject *)__pyx_n_s__genexpr));
   __Pyx_GIVEREF(((PyObject *)__pyx_n_s__genexpr));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_146));
-  __pyx_k_codeobj_147 = (PyObject*)__Pyx_PyCode_New(1, 0, 3, 0, 0, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_k_tuple_146, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_143, __pyx_n_s__decode_sentence, 115, __pyx_empty_bytes); if (unlikely(!__pyx_k_codeobj_147)) {__pyx_filename = __pyx_f[10]; __pyx_lineno = 115; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_145));
+  __pyx_k_codeobj_146 = (PyObject*)__Pyx_PyCode_New(1, 0, 3, 0, 0, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_k_tuple_145, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_142, __pyx_n_s__decode_sentence, 115, __pyx_empty_bytes); if (unlikely(!__pyx_k_codeobj_146)) {__pyx_filename = __pyx_f[10]; __pyx_lineno = 115; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_RefNannyFinishContext();
   return 0;
   __pyx_L1_error:;
@@ -67350,9 +67603,9 @@ 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 = 71; __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 = 72; __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 = 73; __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 = 74; __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;
@@ -67448,7 +67701,7 @@ PyMODINIT_FUNC PyInit__sa(void)
   __pyx_t_2 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__getLogger); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 15; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  __pyx_t_1 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_k_tuple_140), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 15; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_k_tuple_139), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 15; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (PyObject_SetAttr(__pyx_m, __pyx_n_s__logger, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 15; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
@@ -67731,7 +67984,7 @@ PyMODINIT_FUNC PyInit__sa(void)
  * 
  * cdef class TrieNode:
  */
-  __pyx_v_3_sa_EPSILON = __pyx_f_3_sa_sym_fromstring(__pyx_k_148, 1);
+  __pyx_v_3_sa_EPSILON = __pyx_f_3_sa_sym_fromstring(__pyx_k_147, 1);
 
   /* "/Users/vchahun/Sandbox/cdec/python/src/sa/rulefactory.pxi":55
  *     cdef public int count
@@ -67742,7 +67995,7 @@ PyMODINIT_FUNC PyInit__sa(void)
  */
   __pyx_t_1 = __Pyx_PyBool_FromLong(0); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[8]; __pyx_lineno = 55; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __pyx_k_100 = __pyx_t_1;
+  __pyx_k_99 = __pyx_t_1;
   __Pyx_GIVEREF(__pyx_t_1);
   __pyx_t_1 = 0;
 
diff --git a/python/src/sa/data_array.pxi b/python/src/sa/data_array.pxi
index 371708c5..2a8ea921 100644
--- a/python/src/sa/data_array.pxi
+++ b/python/src/sa/data_array.pxi
@@ -7,11 +7,11 @@ from libc.stdlib cimport malloc, realloc, free
 from libc.string cimport memset, strcpy
 
 cdef class DataArray:
-    cdef word2id
-    cdef id2word
+    cdef public word2id
+    cdef public id2word
     cdef public IntList data
     cdef public IntList sent_id
-    cdef IntList sent_index
+    cdef public IntList sent_index
     cdef bint use_sent_id
 
     def __cinit__(self, from_binary=None, from_text=None, side=None, bint use_sent_id=False):
@@ -44,17 +44,18 @@ cdef class DataArray:
             sent.append(self.id2word[self.data.arr[i]])
         return sent
 
-    def get_sentence_position(self, loc):
-        return loc - self.sent_index.arr[self.sent_id.arr[loc]]
-
     def get_id(self, word):
         if not word in self.word2id:
             self.word2id[word] = len(self.id2word)
             self.id2word.append(word)
         return self.word2id[word]
 
-    def get_word(self, id):
-        return self.id2word[id]
+    def __getitem__(self, loc):
+        return self.id2word[self.data.arr[loc]]
+
+    def get_sentence_bounds(self, loc):
+        cdef int sid = self.sent_id.arr[loc]
+        return (self.sent_index.arr[sid], self.sent_index.arr[sid+1])
 
     def write_text(self, char* filename):
         with open(filename, "w") as f:
diff --git a/python/src/sa/suffix_array.pxi b/python/src/sa/suffix_array.pxi
index baa3d546..de4adcd9 100644
--- a/python/src/sa/suffix_array.pxi
+++ b/python/src/sa/suffix_array.pxi
@@ -20,15 +20,6 @@ cdef class SuffixArray:
     def __getitem__(self, i):
         return self.sa.arr[i]
 
-    def get_sentence_id(self, i):
-        return self.darray.get_sentence_id(i)
-
-    def get_sentence(self, i):
-        return self.darray.get_sentence(i)
-
-    def get_sentence_position(self, loc):
-        return self.darray.get_sentence_position(loc)
-
     def read_text(self, filename, side):
         '''Constructs suffix array using the algorithm
         of Larsson & Sadahkane (1999)'''