bug fixing in maca_trans_parser_conll2fann

maca_trans_parser/src/context.c

@@ -333,105 +333,4 @@ context *context_read_options(int argc, char *argv[])
   return ctx;
 }
 
-void context_set_linguistic_resources_filenames_parser(context *ctx)
-{
-  char absolute_path[500];
-  char absolute_filename[500];
-
-  absolute_path[0] = '\0';
-
-  if(ctx->maca_data_path)
-    strcat(absolute_path, ctx->maca_data_path);
-  else
-    strcat(absolute_path, getenv("MACAON_DIR"));
-	   
-  strcat(absolute_path, "/");
-  strcat(absolute_path, ctx->language);
-  strcat(absolute_path, "/bin/");
-
-
-  if(!ctx->perc_model_filename){
-    strcpy(absolute_filename, absolute_path);
-    strcat(absolute_filename, DEFAULT_MODEL_FILENAME);
-    ctx->perc_model_filename = strdup(absolute_filename);
-  }
-
-  if(!ctx->vocabs_filename){
-    strcpy(absolute_filename, absolute_path);
-    strcat(absolute_filename, DEFAULT_VOCABS_FILENAME);
-    ctx->vocabs_filename = strdup(absolute_filename);
-  }
-
-  /*  if(!ctx->mcd_filename){
-    strcpy(absolute_filename, absolute_path);
-    strcat(absolute_filename, DEFAULT_MULTI_COL_DESC_FILENAME);
-    ctx->mcd_filename = strdup(absolute_filename);
-    }*/
-
-  if(!ctx->features_model_filename){
-    strcpy(absolute_filename, absolute_path);
-    strcat(absolute_filename, DEFAULT_FEATURES_MODEL_FILENAME);
-    ctx->features_model_filename = strdup(absolute_filename);
-  }
-
-  fprintf(stderr, "perc_model_filename = %s\n", ctx->perc_model_filename);
-  fprintf(stderr, "vocabs_filename = %s\n", ctx->vocabs_filename);
-  fprintf(stderr, "mcd_filename = %s\n", ctx->mcd_filename);
-  fprintf(stderr, "perc_features_model_filename = %s\n", ctx->features_model_filename);
-}
-
-void context_set_linguistic_resources_filenames_tagger(context *ctx)
-{
-  char absolute_path[500];
-  char absolute_filename[500];
 
-  absolute_path[0] = '\0';
-
-  if(ctx->maca_data_path)
-    strcat(absolute_path, ctx->maca_data_path);
-  else
-    strcat(absolute_path, getenv("MACAON_DIR"));
-	   
-  strcat(absolute_path, "/");
-  strcat(absolute_path, ctx->language);
-  strcat(absolute_path, "/bin/");
-
-  
-  if(!ctx->perc_model_filename){
-    strcpy(absolute_filename, absolute_path);
-    strcat(absolute_filename, DEFAULT_MODEL_TAGGER_FILENAME);
-    ctx->perc_model_filename = strdup(absolute_filename);
-  }
-
-  if(!ctx->vocabs_filename){
-    strcpy(absolute_filename, absolute_path);
-    strcat(absolute_filename, DEFAULT_VOCABS_TAGGER_FILENAME);
-    ctx->vocabs_filename = strdup(absolute_filename);
-  }
-
-  /*  if(!ctx->mcd_filename){
-    strcpy(absolute_filename, absolute_path);
-    strcat(absolute_filename, DEFAULT_MULTI_COL_DESC_TAGGER_FILENAME);
-    ctx->mcd_filename = strdup(absolute_filename);
-    }*/
-
-  if(!ctx->features_model_filename){
-    strcpy(absolute_filename, absolute_path);
-    strcat(absolute_filename, DEFAULT_FEATURES_MODEL_TAGGER_FILENAME);
-    ctx->features_model_filename = strdup(absolute_filename);
-  }
-
-  if(!ctx->f2p_filename){
-    strcpy(absolute_filename, absolute_path);
-    strcat(absolute_filename, DEFAULT_F2P_FILENAME);
-    ctx->f2p_filename = strdup(absolute_filename);
-    ctx->f2p = form2pos_read(ctx->f2p_filename);
-  }
-
-  fprintf(stderr, "perc_model_filename = %s\n", ctx->perc_model_filename);
-  fprintf(stderr, "vocabs_filename = %s\n", ctx->vocabs_filename);
-  fprintf(stderr, "mcd_filename = %s\n", ctx->mcd_filename);
-  fprintf(stderr, "perc_features_model_filename = %s\n", ctx->features_model_filename);
-  fprintf(stderr, "f2p_filename = %s\n", ctx->f2p_filename);
-  
-}

maca_trans_parser/src/context.h

@@ -93,10 +93,4 @@ void context_maca_data_path_help_message(context *ctx);
 void context_f2p_filename_help_message(context *ctx);
 
 
-void context_set_linguistic_resources_filenames_tagger(context *ctx);
-void context_set_linguistic_resources_filenames_parser(context *ctx);
-
-
-
-
 #endif

maca_trans_parser/src/decode.c

@@ -42,6 +42,54 @@ void decode_check_options(context *ctx){
   }
 }
 
+void set_linguistic_resources_filenames_parser(context *ctx)
+{
+  char absolute_path[500];
+  char absolute_filename[500];
+
+  absolute_path[0] = '\0';
+
+  if(ctx->maca_data_path)
+    strcat(absolute_path, ctx->maca_data_path);
+  else
+    strcat(absolute_path, getenv("MACAON_DIR"));
+	   
+  strcat(absolute_path, "/");
+  strcat(absolute_path, ctx->language);
+  strcat(absolute_path, "/bin/");
+
+
+  if(!ctx->perc_model_filename){
+    strcpy(absolute_filename, absolute_path);
+    strcat(absolute_filename, DEFAULT_MODEL_FILENAME);
+    ctx->perc_model_filename = strdup(absolute_filename);
+  }
+
+  if(!ctx->vocabs_filename){
+    strcpy(absolute_filename, absolute_path);
+    strcat(absolute_filename, DEFAULT_VOCABS_FILENAME);
+    ctx->vocabs_filename = strdup(absolute_filename);
+  }
+
+  /*  if(!ctx->mcd_filename){
+    strcpy(absolute_filename, absolute_path);
+    strcat(absolute_filename, DEFAULT_MULTI_COL_DESC_FILENAME);
+    ctx->mcd_filename = strdup(absolute_filename);
+    }*/
+
+  if(!ctx->features_model_filename){
+    strcpy(absolute_filename, absolute_path);
+    strcat(absolute_filename, DEFAULT_FEATURES_MODEL_FILENAME);
+    ctx->features_model_filename = strdup(absolute_filename);
+  }
+
+  fprintf(stderr, "perc_model_filename = %s\n", ctx->perc_model_filename);
+  fprintf(stderr, "vocabs_filename = %s\n", ctx->vocabs_filename);
+  fprintf(stderr, "mcd_filename = %s\n", ctx->mcd_filename);
+  fprintf(stderr, "perc_features_model_filename = %s\n", ctx->features_model_filename);
+}
+
+
 int main(int argc, char *argv[])
 {
   FILE *conll_file = NULL;
@@ -53,7 +101,7 @@ int main(int argc, char *argv[])
   ctx = context_read_options(argc, argv);
   decode_check_options(ctx);
 
-  context_set_linguistic_resources_filenames_parser(ctx);
+  set_linguistic_resources_filenames_parser(ctx);
   ctx->features_model = feat_model_read(ctx->features_model_filename);
 
   ctx->vocabs = dico_vec_read(ctx->vocabs_filename, ctx->hash_ratio);

maca_trans_parser/src/decode_tagger.c

@@ -42,6 +42,64 @@ void decode_check_options(context *ctx){
   }
 }
 
+
+void set_linguistic_resources_filenames_tagger(context *ctx)
+{
+  char absolute_path[500];
+  char absolute_filename[500];
+
+  absolute_path[0] = '\0';
+
+  if(ctx->maca_data_path)
+    strcat(absolute_path, ctx->maca_data_path);
+  else
+    strcat(absolute_path, getenv("MACAON_DIR"));
+	   
+  strcat(absolute_path, "/");
+  strcat(absolute_path, ctx->language);
+  strcat(absolute_path, "/bin/");
+
+  
+  if(!ctx->perc_model_filename){
+    strcpy(absolute_filename, absolute_path);
+    strcat(absolute_filename, DEFAULT_MODEL_TAGGER_FILENAME);
+    ctx->perc_model_filename = strdup(absolute_filename);
+  }
+
+  if(!ctx->vocabs_filename){
+    strcpy(absolute_filename, absolute_path);
+    strcat(absolute_filename, DEFAULT_VOCABS_TAGGER_FILENAME);
+    ctx->vocabs_filename = strdup(absolute_filename);
+  }
+
+  /*  if(!ctx->mcd_filename){
+    strcpy(absolute_filename, absolute_path);
+    strcat(absolute_filename, DEFAULT_MULTI_COL_DESC_TAGGER_FILENAME);
+    ctx->mcd_filename = strdup(absolute_filename);
+    }*/
+
+  if(!ctx->features_model_filename){
+    strcpy(absolute_filename, absolute_path);
+    strcat(absolute_filename, DEFAULT_FEATURES_MODEL_TAGGER_FILENAME);
+    ctx->features_model_filename = strdup(absolute_filename);
+  }
+
+  if(!ctx->f2p_filename){
+    strcpy(absolute_filename, absolute_path);
+    strcat(absolute_filename, DEFAULT_F2P_FILENAME);
+    ctx->f2p_filename = strdup(absolute_filename);
+    ctx->f2p = form2pos_read(ctx->f2p_filename);
+  }
+
+  fprintf(stderr, "perc_model_filename = %s\n", ctx->perc_model_filename);
+  fprintf(stderr, "vocabs_filename = %s\n", ctx->vocabs_filename);
+  fprintf(stderr, "mcd_filename = %s\n", ctx->mcd_filename);
+  fprintf(stderr, "perc_features_model_filename = %s\n", ctx->features_model_filename);
+  fprintf(stderr, "f2p_filename = %s\n", ctx->f2p_filename);
+  
+}
+
+
 int main(int argc, char *argv[])
 {
   FILE *conll_file = NULL;
@@ -50,7 +108,7 @@ int main(int argc, char *argv[])
   ctx = context_read_options(argc, argv);
   decode_check_options(ctx);
 
-  context_set_linguistic_resources_filenames_tagger(ctx);
+  set_linguistic_resources_filenames_tagger(ctx);
   ctx->features_model = feat_model_read(ctx->features_model_filename);
   ctx->vocabs = dico_vec_read(ctx->vocabs_filename, ctx->hash_ratio);
   mcd_link_to_dico(ctx->mcd_struct, ctx->vocabs);

maca_trans_parser/src/maca_trans_parser_conll2fann.c

@@ -27,13 +27,13 @@ void transform_treebank_help_message(context *ctx)
   context_mcd_help_message(ctx);
   context_features_model_help_message(ctx);
   fprintf(stderr, "IN TEST MODE\n");
-  context_alphabet_help_message(ctx);
+  context_vocabs_help_message(ctx);
 
   fprintf(stderr, "OUTPUT\n");
   context_cff_help_message(ctx);
   context_fann_help_message(ctx);
   fprintf(stderr, "IN TRAIN MODE\n");
-  context_alphabet_help_message(ctx);
+  context_vocabs_help_message(ctx);
 
 }
 
@@ -140,6 +140,8 @@ int main(int argc, char *argv[])
   
   ctx = context_read_options(argc, argv);
   transform_treebank_check_options(ctx);
+
+  ctx->features_model = feat_model_read(ctx->features_model_filename);
   
   if(ctx->mode == TRAIN_MODE){
     mcd_extract_dico_from_corpus(ctx->mcd_struct, ctx->conll_filename);
@@ -176,6 +178,15 @@ int main(int argc, char *argv[])
   fprintf(output_file, "%d", nb_trans);
   
   fclose(output_file);
+
+  if(ctx->mode == TRAIN_MODE){
+    /* dico_print(ctx->perceptron_features_filename, ctx->d_perceptron_features); */
+    dico_vec_print(ctx->vocabs_filename, ctx->vocabs);
+    
+  }
+
+
+
   context_free(ctx);
   return 0;
 }