moved word_buffer to maca_common

maca_common/CMakeLists.txt

@@ -7,6 +7,7 @@ set(SOURCES  src/util.c
  src/form2pos.c
  src/word.c
  src/sentence.c
+ src/word_buffer.c
 )
 
 #compiling library

maca_trans_parser/src/word_buffer.h → maca_common/include/word_buffer.h

@@ -5,6 +5,22 @@
 #include"word.h"
 #include"mcd.h"
 
+
+#define word_buffer_get_size(wb) (wb)->size
+#define word_buffer_get_nbelem(wb) (wb)->nbelem
+#define word_buffer_get_lookahead(wb) (wb)->lookahead
+#define word_buffer_get_current_index(wb) (wb)->current_index
+#define word_buffer_get_input_file(wb) (wb)->input_file
+#define word_buffer_get_mcd(wb) (wb)->input_mcd_struct
+
+#define word_buffer_b0(wb) ((wb)->nbelem == 0)? NULL : (wb)->array[(wb)->current_index]
+#define word_buffer_b1(wb) ((wb)->current_index + 1 >= (wb)->nbelem)? NULL : (wb)->array[(wb)->current_index + 1]
+#define word_buffer_b2(wb) ((wb)->current_index + 2 >= (wb)->nbelem)? NULL : (wb)->array[(wb)->current_index + 2]
+#define word_buffer_b3(wb) ((wb)->current_index + 3 >= (wb)->nbelem)? NULL : (wb)->array[(wb)->current_index + 3]
+#define word_buffer_bm1(wb) ((wb)->current_index - 1 < 0)? NULL : (wb)->array[(wb)->current_index - 1]
+#define word_buffer_bm2(wb) ((wb)->current_index - 2 < 0)? NULL : (wb)->array[(wb)->current_index - 2]
+#define word_buffer_bm3(wb) ((wb)->current_index - 3 < 0)? NULL : (wb)->array[(wb)->current_index - 3]
+
 typedef struct {
   int size;           /* size of the array used to store words */
   int nbelem;         /* number of words in the buffer */
@@ -20,6 +36,11 @@ word_buffer *word_buffer_new(FILE *input_file, mcd *mcd_struct, int lookahead);
 void word_buffer_free(word_buffer *wb);
 int word_buffer_add(word_buffer *wb, word *w);
 word *word_buffer_get_word_relative(word_buffer *wb, int dist);
+int word_buffer_read_next_word(word_buffer *wb);
+int word_buffer_move_right(word_buffer *wb);
+int word_buffer_move_left(word_buffer *wb);
+void word_buffer_print(FILE *f, word_buffer *wb);
+/*
 word *word_buffer_b0(word_buffer *wb);
 word *word_buffer_b1(word_buffer *wb);
 word *word_buffer_b2(word_buffer *wb);
@@ -27,9 +48,7 @@ word *word_buffer_b3(word_buffer *wb);
 word *word_buffer_bm1(word_buffer *wb);
 word *word_buffer_bm2(word_buffer *wb);
 word *word_buffer_bm3(word_buffer *wb);
-int word_buffer_read_next_word(word_buffer *wb);
-int word_buffer_move_right(word_buffer *wb);
-int word_buffer_move_left(word_buffer *wb);
-void word_buffer_print(FILE *f, word_buffer *wb);
+*/
+
 
 #endif

maca_trans_parser/src/word_buffer.c → maca_common/src/word_buffer.c

@@ -64,6 +64,41 @@ word *word_buffer_get_word(word_buffer *wb, int offset)
   return ((wb->current_index + offset >=0) && (wb->current_index + offset <= wb->nbelem))? wb->array[wb->current_index + offset] : NULL;
 }
 
+
+int word_buffer_read_next_word(word_buffer *wb)
+{
+  word *w = NULL;
+  int index;
+
+  w = word_read(wb->input_file, wb->mcd_struct);
+  if(w == NULL) return -1;
+
+  
+  index = word_buffer_add(wb, w);  
+  word_set_relative_index(w, index);
+  return index;
+}
+
+int word_buffer_move_right(word_buffer *wb)
+{
+  if((wb->nbelem - 1 - wb->current_index) <= wb->lookahead)
+    word_buffer_read_next_word(wb);
+  if(wb->current_index == wb->nbelem - 1) return 0;
+  wb->current_index++;
+  return 1;
+}
+
+int word_buffer_move_left(word_buffer *wb)
+{
+  if(wb->current_index > 0){
+    wb->current_index--;
+    return 1;
+  }
+  return 0;
+}
+
+
+/*
 word *word_buffer_b0(word_buffer *wb)
 {
   return(wb->nbelem == 0)? NULL : wb->array[wb->current_index];
@@ -98,37 +133,5 @@ word *word_buffer_bm3(word_buffer *wb)
 {
   return(wb->current_index - 3 < 0)? NULL : wb->array[wb->current_index - 3];
 }
-
-int word_buffer_read_next_word(word_buffer *wb)
-{
-  word *w = NULL;
-  int index;
-
-  w = word_read(wb->input_file, wb->mcd_struct);
-  if(w == NULL) return -1;
-
-  
-  index = word_buffer_add(wb, w);  
-  word_set_relative_index(w, index);
-  return index;
-}
-
-int word_buffer_move_right(word_buffer *wb)
-{
-  if((wb->nbelem - 1 - wb->current_index) <= wb->lookahead)
-    word_buffer_read_next_word(wb);
-  if(wb->current_index == wb->nbelem - 1) return 0;
-  wb->current_index++;
-  return 1;
-}
-
-int word_buffer_move_left(word_buffer *wb)
-{
-  if(wb->current_index > 0){
-    wb->current_index--;
-    return 1;
-  }
-  return 0;
-}
-
+*/
 

maca_trans_parser/CMakeLists.txt

@@ -18,7 +18,6 @@ set(SOURCES src/context.c
  src/queue.c
  src/beam.c
  src/feat_types.c
- src/word_buffer.c
 )
 
 #compiling library

maca_trans_parser/src/context.c

@@ -57,7 +57,7 @@ context *context_new(void)
   ctx->f2p_filename = NULL;  
 
   ctx->maca_data_path = NULL;  
-  ctx->language = strdup("fr");
+  ctx->language = strdup("fr_stream");
 
   ctx->root_label = strdup("root");
   ctx->d_perceptron_features = NULL;
@@ -67,7 +67,6 @@ context *context_new(void)
   ctx->dico_labels = NULL;
   ctx->f2p = NULL;
 
-
   ctx->iteration_nb = 4;
   ctx->debug_mode = 0;
   ctx->feature_cutoff = 0;
@@ -77,7 +76,7 @@ context *context_new(void)
   ctx->beam_width = 1;
   ctx->sent_nb = 1000000;
   ctx->hidden_neurons_nb = 100;
-  ctx->stream_mode = 0;
+  ctx->stream_mode = 1;
 
   ctx->conll = 0;
   ctx->ifpls = 1;

maca_trans_parser/src/depset.c

@@ -75,11 +75,21 @@ void depset_print2(FILE *f, depset *d, dico *dico_labels)
 {
   int i;
   int distance;
-  
+  char *label;
   for(i=1; i < d->length; i++){
     if((d->array[i].gov) && (d->array[i].dep)){
 	distance = word_get_relative_index(d->array[i].gov) - word_get_relative_index(d->array[i].dep);
-	fprintf(f, "%s\t%d\t%s\n", d->array[i].dep->input, distance, dico_int2string(dico_labels, d->array[i].label));
+	/* fprintf(f, "%s\t%d\t%s\n", d->array[i].dep->input, distance, dico_int2string(dico_labels, d->array[i].label)); */
+
+	label = dico_int2string(dico_labels, d->array[i].label);
+	fprintf(f, "%s\t%d\t%s\t", d->array[i].dep->input, distance, label);
+	if(!strcmp(label, "eos"))
+	  fprintf(f, "1\n");
+	else
+	  fprintf(f, "0\n");
+	    
+
+	
     }  
   }
 }