modified sentence segmentation by the parser. The sentence segmentation is now...

modified sentence segmentation by the parser. The sentence segmentation is now provoked by a root action

maca_common/include/word_buffer.h

@@ -55,6 +55,12 @@ void         word_buffer_print_compact(FILE *f, word_buffer *wb);
 int          word_buffer_read_sentence(word_buffer *bw);
 word_buffer *word_buffer_load_mcf(char *mcf_filename, mcd *mcd_struct);
 int          word_buffer_locate_token_with_offset(word_buffer *wb, int offset);
+
+word        *word_buffer_get_rightmost_child(word_buffer *wb, word *gov);
+word        *word_buffer_get_rightmost_descendent(word_buffer *wb, word *root);
+
+
+
 /*
 int          word_buffer_is_empty(word_buffer *wb);
 int          word_buffer_is_last(word_buffer *wb);

maca_common/src/word.c

@@ -190,3 +190,4 @@ void word_sprint_col_n(char *s, word *w, int n)
   }
   s[j] = '\0';
 }
+

maca_common/src/word_buffer.c

@@ -187,6 +187,37 @@ int word_buffer_locate_token_with_offset(word_buffer *wb, int offset)
 } 
 
 
+word *word_buffer_get_rightmost_child(word_buffer *wb, word *gov)
+{
+  word *bm1 = word_buffer_bm1(wb);
+  int rightmost_child_index = word_get_index(gov);
+  for(int current_index = word_get_index(gov); current_index <= word_get_index(bm1); current_index++)
+    if(word_get_gov_index(word_buffer_get_word_n(wb, current_index)) == word_get_index(gov))
+      rightmost_child_index = current_index;
+  return word_buffer_get_word_n(wb, rightmost_child_index);
+}
+
+
+word *word_buffer_get_rightmost_descendent(word_buffer *wb, word *root)
+{
+  word *rightmost_descendent = root;
+  word *rightmost_child = NULL;
+  int change = 1;
+
+  while(change){
+    change = 0;
+    rightmost_child = word_buffer_get_rightmost_child(wb, rightmost_descendent);
+    if(word_get_index(rightmost_child) > word_get_index(rightmost_descendent)){
+      rightmost_descendent = rightmost_child;
+      change = 1;
+    }
+  }
+
+  return rightmost_descendent;
+}
+
+
+
 /*int word_buffer_end(word_buffer *wb)
 {
   return (wb->current_index >= wb->nbelem)? 1 : 0;

maca_trans_parser/src/maca_trans_parser_arc_eager_mcf2cff.c

@@ -92,9 +92,6 @@ void generate_training_file(FILE *output_file, context *ctx)
     switch(mvt_type){
     case MVT_PARSER_EOS :
       movement_parser_eos(c);
-      sentence_nb++;
-      if((sentence_nb % 100) == 0)
-	fprintf(stderr, "\rsentence %d", sentence_nb);
       /* if(word_buffer_is_last(ref)) */
       break;
     case MVT_PARSER_LEFT :
@@ -108,6 +105,9 @@ void generate_training_file(FILE *output_file, context *ctx)
       movement_parser_reduce(c);
       break;   
     case MVT_PARSER_ROOT :
+      sentence_nb++;
+      if((sentence_nb % 100) == 0)
+	fprintf(stderr, "\rsentence %d", sentence_nb);
       movement_parser_root(c, root_label);
       break;
     case MVT_PARSER_SHIFT :

maca_trans_parser/src/movements.c

@@ -20,7 +20,7 @@ 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;
+  //  gov->lspan = dep->lspan;
   
   config_push_mvt(c, movement_code, gov, dep);
 
@@ -57,7 +57,7 @@ int movement_right_arc(config *c, int movement_code, int label)
   word_set_gov(dep, dist);
   word_set_label(dep, label);
 
-  gov->rspan = dep->rspan;
+  //  gov->rspan = dep->rspan;
 
   
   stack_push(config_get_stack(c), word_buffer_b0(config_get_buffer(c)));
@@ -139,6 +139,30 @@ int movement_root(config *c, int movement_code, int root_code)
   word_set_label(s0, root_code);
   s0->is_root = 1;
 
+  /* the word bm1 (preceding the current word in the buffer) is marked as end of sentence */ 
+  word *bm1 = word_buffer_bm1(config_get_buffer(c));
+  if(bm1)
+  word_set_sent_seg(bm1, 1); 
+ 
+  
+  /*  word *rd = word_buffer_get_rightmost_descendent(config_get_buffer(c), s0);
+  if(rd)
+  word_set_sent_seg(rd, 1); */
+
+  stack_pop(config_get_stack(c));
+
+  config_push_mvt(c, movement_code, s0, NULL);
+  return 1;
+}
+
+int movement_root_old(config *c, int movement_code, int root_code)
+{
+  if(stack_is_empty(config_get_stack(c))) return 0;
+  word *s0 = stack_top(config_get_stack(c));
+  word_set_gov(s0, 0);
+  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); 
   /*-----------------------------*/

maca_trans_parser/src/oracle_parser_arc_eager.c

@@ -69,12 +69,12 @@ 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)
        && check_all_dependents_of_word_in_ref_are_in_hyp(c, ref, s0_index)
        ){
       return MVT_PARSER_EOS;
-    }
+      }*/
 
     /* LEFT ARC  b0 is the governor and s0 the dependent */
     if(s0_gov_index == b0_index){

maca_trans_parser/src/simple_decoder_parser_arc_eager.c

@@ -134,6 +134,7 @@ void simple_decoder_parser_arc_eager(context *ctx)
   int argmax1, argmax2;
   float max1, max2;
   int index;
+  word *root;
   
   root_label = dico_string2int(ctx->dico_labels, ctx->root_label);
   if(root_label == -1) root_label = 0;
@@ -212,9 +213,15 @@ void simple_decoder_parser_arc_eager(context *ctx)
 	result = movement_parser_reduce(c);
 	break;
       case MVT_PARSER_ROOT:
+	root = stack_top(config_get_stack(c));
 	result = movement_parser_root(c, root_label);
-	/*	while(!stack_is_empty(config_get_stack(c)))
-		movement_parser_root(c, root_label);*/
+	while(!stack_is_empty(config_get_stack(c))){
+	  /*	  word *dep = stack_pop(config_get_stack(c));
+	  int dist = (word_get_index(root)) - (word_get_index(dep));
+	  word_set_gov(dep, dist);
+	  word_set_label(dep, 0);*/
+	  movement_parser_root(c, root_label);
+	}
 	break;
       case MVT_PARSER_EOS:
 	result = movement_parser_eos(c);