added prediction.c in maca_morpho

maca_morpho/CMakeLists.txt

 set(SOURCES
   src/feat_fct.c
   src/context.c
+  src/vectorize.c
 )
 
 
@@ -21,3 +22,9 @@ target_link_libraries(fplm2cff maca_common)
 target_link_libraries(fplm2cff maca_morpho)
 install (TARGETS fplm2cff DESTINATION bin)
 
+add_executable(predict ./src/predict.c)
+target_link_libraries(predict perceptron)
+target_link_libraries(predict maca_common)
+target_link_libraries(predict maca_morpho)
+install (TARGETS predict DESTINATION bin)
+

maca_morpho/src/context.c

@@ -13,6 +13,7 @@ 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);
   free(ctx);
@@ -30,6 +31,7 @@ context *context_new(void)
   ctx->language = strdup("fr");
   ctx->maca_data_path = NULL;
   ctx->features_filename = NULL;
+  ctx->cfw_filename = NULL;
   return ctx;
 }
 
@@ -62,6 +64,13 @@ 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");
+}
 
 context *context_read_options(int argc, char *argv[])
 {
@@ -71,7 +80,7 @@ context *context_read_options(int argc, char *argv[])
 
   ctx->program_name = strdup(argv[0]);
 
-  static struct option long_options[9] =
+  static struct option long_options[10] =
     {
       {"help",                no_argument,       0, 'h'},
       {"verbose",             no_argument,       0, 'v'},
@@ -81,12 +90,13 @@ context *context_read_options(int argc, char *argv[])
       {"fplm",                required_argument, 0, 'f'},
       {"maca_data_path",      required_argument, 0, 'D'},
       {"fm",                  required_argument, 0, 'F'},
-      {"feat",                required_argument, 0, 'x'}
+      {"feat",                required_argument, 0, 'x'},
+      {"weights",             required_argument, 0, 'w'}
     };
   optind = 0;
   opterr = 0;
   
-  while ((c = getopt_long (argc, argv, "hvdf:L:M:D:F:x:", long_options, &option_index)) != -1){ 
+  while ((c = getopt_long (argc, argv, "hvdf:L:M:D:F:x:w:", long_options, &option_index)) != -1){ 
     switch (c)
       {
       case 'd':
@@ -113,6 +123,9 @@ context *context_read_options(int argc, char *argv[])
       case 'x':
 	ctx->features_filename = strdup(optarg);
 	break;
+      case 'w':
+	ctx->cfw_filename = strdup(optarg);
+	break;
       }
   }
 

maca_morpho/src/context.h

@@ -18,6 +18,7 @@ typedef struct {
   char *maca_data_path;
   char *fm_filename;
   char *features_filename;
+  char *cfw_filename;
 } context;
 
 
@@ -31,4 +32,6 @@ void context_language_help_message(context *ctx);
 void     context_fplm_help_message(context *ctx);
 void     context_maca_data_path_help_message(context *ctx);
 void     context_features_filename_help_message(context *ctx);
+void     context_weights_matrix_filename_help_message(context *ctx);
+void     context_features_model_help_message(context *ctx);
 #endif

maca_morpho/src/fplm2cff.c

@@ -7,43 +7,7 @@
 #include "feat_vec.h"
 #include "dico.h"
 #include "util.h"
-
-#define LOOKUP_MODE 1
-#define TRAIN_MODE 2
-
-int get_feat_value(feat_model *fm, char *form, dico *dico_features, int feat_nb, int mode)
-{
-  feat_desc *fd = fm->array[feat_nb];
-  int i;
-  int feat_val;
-  char str[10];
-
-  /*  the name of the feature is built in fm->string and its value in the dictionnary (dico_features) is returned */
-  fm->string[0] = '\0';
-  for(i=0; i < fd->nbelem; i++){
-    strcat(fm->string, fd->array[i]->name);
-    feat_val = fd->array[i]->fct(form);
-    sprintf(str, "%d", feat_val);
-    strcat(fm->string, str);
-
-    /* catenate_int(fm->string, feat_val); */
-  }
-  if(mode == LOOKUP_MODE){
-    if(fm->string)
-    return dico_string2int(dico_features, fm->string);
-  } 
-  return dico_add(dico_features, fm->string);
-}
-
-
-feat_vec *form2fv(char *form, feat_vec *fv, feat_model *fm, dico *dico_features)
-{
-  int i;
-  feat_vec_empty(fv);
-  for(i=0; i < fm->nbelem; i++)
-    feat_vec_add(fv, get_feat_value(fm, form, dico_features, i, TRAIN_MODE));
-  return fv;
-}
+#include "vectorize.h"
 
 
 int main(int argc, char *argv[])
@@ -55,6 +19,7 @@ int main(int argc, char *argv[])
     context_fplm_help_message(ctx);
     context_maca_data_path_help_message(ctx);
     context_features_filename_help_message(ctx);
+    context_features_model_help_message(ctx);
     exit(1);
   }
   feat_vec *fv = feat_vec_new(10);
@@ -64,7 +29,7 @@ int main(int argc, char *argv[])
   while(strcmp(form, "end")){
     fscanf(stdin, "%s", form);
     printf("form = %s\n", form);
-    form2fv(form, fv, fm, dico_features);
+    form2fv(form, fv, fm, dico_features, ADD_MODE);
 /* void      feat_vec_print_string(feat_vec *fv, dico *dico_features); */
     feat_vec_print(stdout, fv);
   }

maca_morpho/src/predict.c

+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "context.h"
+#include "feat_model.h"
+#include "feat_vec.h"
+#include "dico.h"
+#include "util.h"
+#include "vectorize.h"
+#include "feature_table.h"
+
+void predict_help_message(context *ctx)
+{
+  context_general_help_message(ctx);
+  context_language_help_message(ctx);
+  context_fplm_help_message(ctx);
+  context_maca_data_path_help_message(ctx);
+  context_features_filename_help_message(ctx);    
+  context_weights_matrix_filename_help_message(ctx);
+  context_features_model_help_message(ctx);
+  exit(1);
+}
+
+
+int main(int argc, char *argv[])
+{
+  context *ctx = context_read_options(argc, argv);
+  if(ctx->help) predict_help_message(ctx);
+  feature_table *cfw = feature_table_load(ctx->cfw_filename, ctx->verbose);
+  feat_vec *fv = feat_vec_new(10);
+  dico *dico_features = dico_read(ctx->features_filename, 0.5);
+  feat_model *fm = feat_model_read(ctx->fm_filename, feat_lib_build(), ctx->verbose); 
+  char form[100];
+  int class;
+  float max;
+  
+  while(strcmp(form, "end")){
+    fscanf(stdin, "%s", form);
+    printf("form = %s\n", form);
+    form2fv(form, fv, fm, dico_features, LOOKUP_MODE);
+    class = feature_table_argmax(fv, cfw, &max);
+    feat_vec_print(stdout, fv);
+    printf("class = %d\n", class);
+    
+  }
+
+  if(ctx->features_filename)
+    dico_print(ctx->features_filename, dico_features);
+
+
+  
+}

maca_morpho/src/vectorize.c

+#include<stdio.h>
+#include<stdlib.h>
+#include<string.h>
+#include"vectorize.h"
+
+int get_feat_value(feat_model *fm, char *form, dico *dico_features, int feat_nb, int mode)
+{
+  feat_desc *fd = fm->array[feat_nb];
+  int i;
+  int feat_val;
+  char str[10];
+
+  /*  the name of the feature is built in fm->string and its value in the dictionnary (dico_features) is returned */
+  fm->string[0] = '\0';
+  for(i=0; i < fd->nbelem; i++){
+    strcat(fm->string, fd->array[i]->name);
+    feat_val = fd->array[i]->fct(form);
+    sprintf(str, "%d", feat_val);
+    strcat(fm->string, str);
+
+    /* catenate_int(fm->string, feat_val); */
+  }
+  if(mode == LOOKUP_MODE){
+    if(fm->string)
+    return dico_string2int(dico_features, fm->string);
+  } 
+  return dico_add(dico_features, fm->string);
+}
+
+
+feat_vec *form2fv(char *form, feat_vec *fv, feat_model *fm, dico *dico_features, int mode)
+{
+  int i;
+  feat_vec_empty(fv);
+  for(i=0; i < fm->nbelem; i++)
+    feat_vec_add(fv, get_feat_value(fm, form, dico_features, i, mode));
+  return fv;
+}

maca_morpho/src/vectorize.h

+#ifndef __VECTORIZE__
+#define __VECTORIZE__
+
+#include"dico.h"
+#include"feat_model.h"
+#include"feat_vec.h"
+
+#define LOOKUP_MODE 1
+#define ADD_MODE 2
+
+
+feat_vec *form2fv(char *form, feat_vec *fv, feat_model *fm, dico *dico_features, int mode);
+
+#endif