modified maca_lexer so that it can take as input speakers and word duration (very ad hoc)!

maca_lexer/src/maca_lexer.c

@@ -38,7 +38,7 @@ int look_for_accept_state_in_path(trie *mwe_trie, int *states_array, int path_in
   return -1;
 }
 
-void print_states_array(char *buffer, context *ctx, trie *mwe_trie, dico *d_mwe_tokens, int *states_array, int *symbols_array, int path_index)
+void print_states_array(char *buffer, context *ctx, trie *mwe_trie, dico *d_mwe_tokens, int *states_array, int *symbols_array, int path_index, float *start_array, float *end_array, int orfeo, char *spkr)
 {
   int i;
   if(path_index == 0) return;
@@ -54,18 +54,30 @@ void print_states_array(char *buffer, context *ctx, trie *mwe_trie, dico *d_mwe_
       else printf("%s\t0\n", dico_int2string(d_mwe_tokens, symbols_array[i]));
     }
   }
+  if(i>0 && orfeo){
+    printf("\t%f\t%f\t%s", start_array[0], end_array[accept_state_index], spkr);
+  }
+      
+
   if(ctx->paste)
     if(accept_state_index != -1) printf("\n"); 
   /* all tokens in path s.t. accept_state_index < token_index < path_index do not form an mwe, they are just printed */
   for(i = accept_state_index + 1; i < path_index; i++){
-    if(ctx->paste)
+    if(ctx->paste){
+      if(orfeo){
+	printf("%s\t%f\t%f\t%s\n", dico_int2string(d_mwe_tokens, symbols_array[i]), start_array[i], end_array[i], spkr);
+      }
+      else{
 	printf("%s\n", dico_int2string(d_mwe_tokens, symbols_array[i]));
+      }
+    }
     else
       printf("%s\t1\n", dico_int2string(d_mwe_tokens, symbols_array[i]));
   }
 }
 
 
+
 int main(int argc, char *argv[])
 {
   char buffer[10000];
@@ -77,9 +89,15 @@ int main(int argc, char *argv[])
   dico *d_mwe_tokens = NULL;
   int states_array[100]; /* an array in which we store the states we have traversed in the trie */
   int symbols_array[100];
+  float start_array[100];
+  float end_array[100];
   int path_index = 0;
   int next_state;
-  
+  int orfeo = 1;
+  char form[1000];
+  float start;
+  float end;
+  char spkr[1000];
   ctx = context_read_options(argc, argv);
    maca_lexer_check_options(ctx);
    
@@ -109,16 +127,29 @@ int main(int argc, char *argv[])
       continue;
     }
     buffer[strlen(buffer)-1] = '\0';
+
+    if(orfeo){
+      sscanf(buffer,"%s\t%f\t%f\t%s", form, &start, &end, spkr);
+      /* look for code of form read */
+      form_code = dico_string2int(d_mwe_tokens, form);
+      
+    }
+    else{
       /* look for code of form read */
       form_code = dico_string2int(d_mwe_tokens, buffer);
+    }
 
     if(form_code == -1){
     /* if form has no code, it cannot be part of a mwe, print the potential mwe discovered so far */
-      print_states_array(buffer, ctx, mwe_trie, d_mwe_tokens, states_array, symbols_array, path_index);
+      print_states_array(buffer, ctx, mwe_trie, d_mwe_tokens, states_array, symbols_array, path_index, start_array, end_array, orfeo, spkr);
       path_index = 0;
       /* print the current form */
-      if(ctx->paste)
+      if(ctx->paste){
+	if(orfeo)
+	  printf("%s\t%f\t%f\t%s\n", form, start, end, spkr);
+	else
 	  printf("%s\n", buffer);
+      }
       else
 	printf("%s\t1\n", buffer);
       continue;
@@ -130,12 +161,16 @@ int main(int argc, char *argv[])
     if(next_state != 0){
       /* the path is growing */
       symbols_array[path_index] = form_code;
+      if(orfeo){
+	start_array[path_index] = start;
+	end_array[path_index] = end;
+      }
       states_array[path_index] = next_state;
       path_index++;
       continue;
     }
     /* print the potential mwe discovered so far */
-    print_states_array(buffer, ctx, mwe_trie, d_mwe_tokens, states_array, symbols_array, path_index);
+    print_states_array(buffer, ctx, mwe_trie, d_mwe_tokens, states_array, symbols_array, path_index, start_array, end_array, orfeo, spkr);
 
     if(path_index != 0)
       /* if there was a path that aborted, see if there is a valid transition from state 0 with form */
@@ -145,14 +180,22 @@ int main(int argc, char *argv[])
     if(next_state){
       /* such a transition exists */
       symbols_array[path_index] = form_code;
+      if(orfeo){
+	start_array[path_index] = start;
+	end_array[path_index] = end;
+      }
       states_array[path_index] = next_state;
       path_index++;
       continue;
     }
     
     /* such a transition does not exist, just print the form */
-    if(ctx->paste)
+    if(ctx->paste){
+      if(orfeo)
+	printf("%s\t%f\t%f\t%s\n", form, start, end, spkr);
+      else
 	printf("%s\n", buffer);
+    }
     else
       printf("%s\t1\n", buffer);
   }
@@ -160,7 +203,7 @@ int main(int argc, char *argv[])
   if(path_index != 0){
     /* there is something in states array */
     /* print the potential mwe discovered so far */
-    print_states_array(buffer, ctx, mwe_trie, d_mwe_tokens, states_array, symbols_array, path_index);
+    print_states_array(buffer, ctx, mwe_trie, d_mwe_tokens, states_array, symbols_array, path_index, start_array, end_array, orfeo, spkr);
     path_index = 0;
   }