generate train and test files, generate cff, predict test's forms' classes

maca_morpho/src/maca_morpho_context.c

+#include<stdlib.h>
+#include<stdio.h>
+#include<string.h>
+#include<unistd.h>
+#include<getopt.h>
+#include "maca_morpho_context.h"
+#include "util.h"
+
+
+void context_set_linguistic_resources_filenames(context *ctx);
+
+void context_free(context *ctx)
+{
+  if(ctx->program_name) free(ctx->program_name);
+  if(ctx->fplm_filename) free(ctx->fplm_filename);
+  if(ctx->cfw_filename) free(ctx->cfw_filename);
+  if(ctx->language) free(ctx->language);
+  if(ctx->maca_data_path) free(ctx->maca_data_path);
+  if(ctx->class_name)	free(ctx->class_name);
+  free(ctx);
+}
+
+context *context_new(void)
+{
+  context *ctx = (context *)memalloc(sizeof(context));
+
+  ctx->help = 0;
+  ctx->verbose = 0;
+  ctx->debug_mode = 0;
+  ctx->program_name = NULL;
+  ctx->fplm_filename = NULL;
+  ctx->language = strdup("fr");
+  ctx->maca_data_path = NULL;
+  ctx->features_filename = NULL;
+  ctx->cfw_filename = NULL;
+  ctx->class_name = NULL;
+  return ctx;
+}
+
+void context_general_help_message(context *ctx)
+{
+    fprintf(stderr, "usage: %s [options]\n", ctx->program_name);
+    fprintf(stderr, "Options:\n");
+    fprintf(stderr, "\t-h --help             : print this message\n");
+    fprintf(stderr, "\t-v --verbose          : activate verbose mode\n");
+    fprintf(stderr, "\t-r --hratio   <float> : set the occupation ratio of hash tables (default is 0.5)\n");
+}
+
+void context_fplm_help_message(context *ctx){
+  fprintf(stderr, "\t-f --fplm   <file>  : fplm (form pos lemma morpho) file\n");
+}
+
+void context_language_help_message(context *ctx){
+  fprintf(stderr, "\t-L --language  : identifier of the language to use\n");
+}
+
+void context_maca_data_path_help_message(context *ctx){
+  fprintf(stderr, "\t-M --maca_data_path  : path to maca_data directory\n");
+}
+
+void context_fm_help_message(context *ctx){
+  fprintf(stderr, "\t-F --fm <file> : feature model file name\n");
+}
+
+void context_features_filename_help_message(context *ctx){
+  fprintf(stderr, "\t-x --feat <file> : features dictionary file name\n");
+}
+
+void context_weights_matrix_filename_help_message(context *ctx){
+  fprintf(stderr, "\t-w --weights <file> : weight matrix (cfw) filename\n");
+}
+
+void context_features_model_help_message(context *ctx){
+  fprintf(stderr, "\t-F --feat_model <file> : feature model file name\n");
+}
+
+void context_class_help_message(context *ctx){
+  fprintf(stderr, "\t-c --class <string>  : tense, person, gender or number\n");
+}
+
+void context_fplm_test_percent_help_message(context *ctx){
+  fprintf(stderr, "\t-p --percent <int>  : percentage of the fplm file to make a test file\n");
+}
+
+context *context_read_options(int argc, char *argv[])
+{
+  int c;
+  int option_index = 0;
+  context *ctx = context_new();
+
+  ctx->program_name = strdup(argv[0]);
+
+  static struct option long_options[12] =
+    {
+      {"help",                no_argument,       0, 'h'},
+      {"verbose",             no_argument,       0, 'v'},
+      {"debug",               no_argument,       0, 'd'},
+      {"mcd",                 required_argument, 0, 'C'}, 
+      {"language",            required_argument, 0, 'L'},
+      {"fplm",                required_argument, 0, 'f'},
+      {"maca_data_path",      required_argument, 0, 'D'},
+      {"fm",                  required_argument, 0, 'F'},
+      {"feat",                required_argument, 0, 'x'},
+      {"weights",             required_argument, 0, 'w'},
+      {"class", 			  required_argument, 0, 'c'},
+      {"percent",             required_argument, 0, 'p'}
+    };
+  optind = 0;
+  opterr = 0;
+  
+  while ((c = getopt_long (argc, argv, "hvdf:L:M:D:F:x:w:c:p:", long_options, &option_index)) != -1){ 
+    switch (c)
+      {
+      case 'd':
+	ctx->debug_mode = 1;
+	break;
+      case 'h':
+	ctx->help = 1;
+	break;
+      case 'v':
+	ctx->verbose = 1;
+	break;
+      case 'f':
+	ctx->fplm_filename = strdup(optarg);
+	break;
+      case 'L':
+	ctx->language = strdup(optarg);
+	break;
+      case 'D':
+	ctx->maca_data_path = strdup(optarg);
+	break;
+      case 'F':
+	ctx->fm_filename = strdup(optarg);
+	break;
+      case 'x':
+	ctx->features_filename = strdup(optarg);
+	break;
+      case 'w':
+	ctx->cfw_filename = strdup(optarg);
+	break;
+		case 'c':
+			ctx->class_name = strdup(optarg);
+			break;
+		case 'p':
+			ctx->fplm_test_percent = atoi(optarg);
+			break;
+      }
+  }
+
+  context_set_linguistic_resources_filenames(ctx);
+
+  return ctx;
+}
+
+void context_set_linguistic_resources_filenames(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 {
+      char *e = getenv("MACAON_DIR");
+      if (e != NULL) {
+	  strcat(absolute_path, e);	  
+      } else {
+	  fprintf(stderr, "ATTENTION: the environment variable MACAON_DIR is not defined\n");
+      }
+  }
+
+	   
+  strcat(absolute_path, "/");
+  strcat(absolute_path, ctx->language);
+  strcat(absolute_path, "/bin/");
+
+  if(!ctx->fplm_filename){
+    strcpy(absolute_filename, absolute_path);
+    strcat(absolute_filename, DEFAULT_FPLM_FILENAME);
+    ctx->fplm_filename = strdup(absolute_filename);
+  }
+  
+}