From b662e2f8a4d35b37bfee80a84e0339e8a873c6af Mon Sep 17 00:00:00 2001
From: Alexis Nasr <alexis.nasr@lif.univ-mrs.fr>
Date: Tue, 19 Dec 2017 22:01:22 +0100
Subject: [PATCH] simplified the actions system, got rid of action EOS, action
ROOT is now responsible to set the sentence boundary
---
maca_common/include/word.h | 2 ++
maca_common/src/word_buffer.c | 2 +-
maca_trans_parser/src/movement_parser_arc_eager.h | 6 +++---
maca_trans_parser/src/movements.c | 8 ++++++++
maca_trans_parser/src/oracle_parser_arc_eager.c | 4 ++--
5 files changed, 16 insertions(+), 6 deletions(-)
diff --git a/maca_common/include/word.h b/maca_common/include/word.h
index 33fd8c1..c6a4b56 100644
--- a/maca_common/include/word.h
+++ b/maca_common/include/word.h
@@ -16,6 +16,8 @@ typedef struct _word {
char16 *form_char16;
int index;
int is_root;
+ int lspan;
+ int rspan;
} word;
/*
diff --git a/maca_common/src/word_buffer.c b/maca_common/src/word_buffer.c
index e9355d5..b72f72e 100644
--- a/maca_common/src/word_buffer.c
+++ b/maca_common/src/word_buffer.c
@@ -80,7 +80,7 @@ int word_buffer_add(word_buffer *wb, word *w)
}
wb->array[wb->nbelem] = w;
word_set_index(w, wb->nbelem);
-
+ w->rspan = w->lspan = wb->nbelem;
wb->nbelem++;
return wb->nbelem - 1;
}
diff --git a/maca_trans_parser/src/movement_parser_arc_eager.h b/maca_trans_parser/src/movement_parser_arc_eager.h
index 2b11bf0..3a040d5 100644
--- a/maca_trans_parser/src/movement_parser_arc_eager.h
+++ b/maca_trans_parser/src/movement_parser_arc_eager.h
@@ -7,9 +7,9 @@
#define MVT_PARSER_SHIFT 0
#define MVT_PARSER_REDUCE 1
#define MVT_PARSER_ROOT 2
-#define MVT_PARSER_EOS 3
-#define MVT_PARSER_LEFT 4
-#define MVT_PARSER_RIGHT 5
+#define MVT_PARSER_EOS -1
+#define MVT_PARSER_LEFT 3
+#define MVT_PARSER_RIGHT 4
/* even movements are left movements (except 0, which is shift and 2 which is root) */
#define movement_parser_left_code(label) (2 * (label) + 4)
diff --git a/maca_trans_parser/src/movements.c b/maca_trans_parser/src/movements.c
index 36e3ff7..8622e4a 100644
--- a/maca_trans_parser/src/movements.c
+++ b/maca_trans_parser/src/movements.c
@@ -47,6 +47,8 @@ int movement_left_arc(config *c, int movement_code, int label)
word_set_gov(dep, dist);
word_set_label(dep, label);
+ gov->lspan = dep->lspan;
+
config_push_mvt(c, movement_code, gov, dep);
stack_pop(config_get_stack(c));
@@ -81,6 +83,9 @@ int movement_right_arc(config *c, int movement_code, int label)
/* create a new dependency */
word_set_gov(dep, dist);
word_set_label(dep, label);
+
+ gov->rspan = dep->rspan;
+
stack_push(config_get_stack(c), word_buffer_b0(config_get_buffer(c)));
word_buffer_move_right(config_get_buffer(c));
@@ -152,6 +157,9 @@ int movement_root(config *c, int movement_code, int root_code)
word_set_label(s0, root_code);
s0->is_root = 1;
+ /*---- ajout 19/12/17 ----------*/
+ word_set_sent_seg(word_buffer_get_word_n(config_get_buffer(c), s0->rspan), 1);
+ /*-----------------------------*/
stack_pop(config_get_stack(c));
config_push_mvt(c, movement_code, s0, NULL);
diff --git a/maca_trans_parser/src/oracle_parser_arc_eager.c b/maca_trans_parser/src/oracle_parser_arc_eager.c
index 074400e..e258c93 100644
--- a/maca_trans_parser/src/oracle_parser_arc_eager.c
+++ b/maca_trans_parser/src/oracle_parser_arc_eager.c
@@ -68,10 +68,10 @@ int oracle_parser_arc_eager(config *c, word_buffer *ref, int root_label)
}
/* word on the top of the stack is an end of sentence marker */
- if((word_get_sent_seg(word_buffer_get_word_n(ref, s0_index)) == 1)
+ /* if((word_get_sent_seg(word_buffer_get_word_n(ref, s0_index)) == 1)
&& (word_get_sent_seg(word_buffer_get_word_n(config_get_buffer(c), s0_index)) != 1)){
return MVT_PARSER_EOS;
- }
+ }*/
/* LEFT ARC b0 is the governor and s0 the dependent */
if(s0_gov_index == b0_index){
--
GitLab