extracted perceptron from maca_trans_parser it is now completely independent

CMakeLists.txt

@@ -4,8 +4,10 @@ project(macaon2)
 add_definitions("-Wall")
 
 include_directories(maca_common/include)
+include_directories(perceptron/lib/include)
 
 add_subdirectory(maca_common)
+add_subdirectory(perceptron)
 add_subdirectory(maca_lemmatizer)
 add_subdirectory(maca_trans_parser)
 add_subdirectory(maca_crf_tagger)

maca_trans_parser/CMakeLists.txt

 set(SOURCES src/context.c
  src/feat_desc.c
- src/feature_table.c
  src/movement_parser.c
  src/movement_tagger.c
  src/feat_fct.c
- src/feat_vec.c
  src/global_feat_vec.c
  src/oracle_parser.c
  src/oracle_tagger.c
  src/simple_decoder_parser.c
  src/simple_decoder_forrest.c
  src/simple_decoder_tagger.c
- src/cf_file.c
  src/feat_lib.c
- src/perceptron.c
  src/stack.c
  src/config2feat_vec.c
  src/depset.c
@@ -27,61 +23,53 @@ set(SOURCES src/context.c
 #compiling library
 include_directories(src)
 add_library(transparse STATIC ${SOURCES})
+target_link_libraries(transparse perceptron)
 
 #compiling, linking and installing executables
 
 add_executable(maca_trans_tagger_mcf2cff ./src/maca_trans_tagger_mcf2cff.c)
+target_link_libraries(maca_trans_tagger_mcf2cff perceptron)
 target_link_libraries(maca_trans_tagger_mcf2cff transparse)
 target_link_libraries(maca_trans_tagger_mcf2cff maca_common)
 install (TARGETS maca_trans_tagger_mcf2cff DESTINATION bin)
 
 add_executable(maca_trans_parser_mcf2fann ./src/maca_trans_parser_mcf2fann.c)
+target_link_libraries(maca_trans_parser_mcf2fann perceptron)
 target_link_libraries(maca_trans_parser_mcf2fann transparse)
 target_link_libraries(maca_trans_parser_mcf2fann maca_common)
 install (TARGETS maca_trans_parser_mcf2fann DESTINATION bin)
 
 add_executable(maca_trans_parser_mcf2cff ./src/maca_trans_parser_mcf2cff.c)
+target_link_libraries(maca_trans_parser_mcf2cff perceptron)
 target_link_libraries(maca_trans_parser_mcf2cff transparse)
 target_link_libraries(maca_trans_parser_mcf2cff maca_common)
 install (TARGETS maca_trans_parser_mcf2cff DESTINATION bin)
 
 add_executable(maca_trans_parser ./src/maca_trans_parser.c)
+target_link_libraries(maca_trans_parser perceptron)
 target_link_libraries(maca_trans_parser transparse)
 target_link_libraries(maca_trans_parser maca_common)
 install (TARGETS maca_trans_parser DESTINATION bin)
 
 add_executable(maca_trans_parser_forrest ./src/decode_forrest.c)
+target_link_libraries(maca_trans_parser_forrest perceptron)
 target_link_libraries(maca_trans_parser_forrest transparse)
 target_link_libraries(maca_trans_parser_forrest maca_common)
 install (TARGETS maca_trans_parser_forrest DESTINATION bin)
 
 add_executable(maca_trans_tagger ./src/maca_trans_tagger.c)
+target_link_libraries(maca_trans_tagger perceptron)
 target_link_libraries(maca_trans_tagger transparse)
 target_link_libraries(maca_trans_tagger maca_common)
 install (TARGETS maca_trans_tagger DESTINATION bin)
 
 add_executable(maca_trans_parser_train ./src/train_perceptron.c)
 target_compile_options(maca_trans_parser_train INTERFACE -Wall)
+target_link_libraries(maca_trans_parser_train perceptron)
 target_link_libraries(maca_trans_parser_train transparse)
 target_link_libraries(maca_trans_parser_train maca_common)
 install (TARGETS maca_trans_parser_train DESTINATION bin)
 
-add_executable(maca_trans_parser_train_from_cff ./src/train.c)
-target_link_libraries(maca_trans_parser_train_from_cff transparse)
-target_link_libraries(maca_trans_parser_train_from_cff maca_common)
-install (TARGETS maca_trans_parser_train_from_cff DESTINATION bin)
-
-add_executable(maca_trans_parser_cff_cutoff ./src/cff_cutoff.c)
-target_link_libraries(maca_trans_parser_cff_cutoff transparse)
-target_link_libraries(maca_trans_parser_cff_cutoff maca_common)
-install (TARGETS maca_trans_parser_cff_cutoff DESTINATION bin)
-
-add_executable(eval_classifier ./src/eval_classifier.c)
-target_link_libraries(eval_classifier transparse)
-target_link_libraries(eval_classifier maca_common)
-install (TARGETS eval_classifier DESTINATION bin)
-
-
 
 #add_executable(test_w2v ./src/test_w2v.c)
 #target_link_libraries(test_w2v transparse)

maca_trans_parser/src/maca_trans_parser_mcf2fann.c

@@ -7,6 +7,7 @@
 #include"oracle_parser.h"
 #include"feat_fct.h"
 #include"context.h"
+#include"feat_types.h"
 #include"feat_vec.h"
 #include"dico_vec.h"
 #include"word_emb.h"
@@ -169,3 +170,94 @@ int main(int argc, char *argv[])
   return 0;
 }
 
+
+
+
+void feat_vec_print_dnn(FILE *f, feat_vec *fv, feat_model *fm, mcd *m)
+{
+  int i,j;
+  feat_desc *fd;
+  simple_feat_desc *sfd;
+  int mcd_representation;
+  int column;
+
+  for(i=0; i < fm->nbelem; i++){
+    fd = fm->array[i];
+    sfd = fd->array[0];
+    
+    if((sfd->type <= FEAT_TYPE_INT_16) && (sfd->type >= FEAT_TYPE_INT_1)){
+      int range = sfd->type - FEAT_TYPE_INT_0;
+      for(j=0; j < range; j++){
+	/* if(i + j != 0) fprintf(f, " "); */
+	if(j == fv->t[i])
+	  fprintf(f, "1 ");
+	else
+	  fprintf(f, "0 ");
+      }
+    }
+    else{
+      column = m->wf2col[sfd->type];
+      mcd_representation = m->representation[column];
+      if(mcd_representation == MCD_REPRESENTATION_EMB){
+	/* if(i != 0) fprintf(f, " "); */
+	word_emb_print(f, m->word_emb_array[column], fv->t[i]);
+      }
+      else if(mcd_representation == MCD_REPRESENTATION_VOCAB){
+	for(j=0; j < m->dico_array[column]->nbelem; j++){
+	  /* if(i + j != 0) fprintf(f, " "); */
+	  if(j == fv->t[i])
+	    fprintf(f, "1 ");
+	  else
+	    fprintf(f, "0 ");
+	}
+      }
+    }
+  fprintf(f, "\n");
+
+  }
+  fprintf(f, "\n");
+}
+
+/* void feat_vec_fill_input_array_dnn(fann_type *input_array, feat_vec *fv, feat_model *fm, mcd *m) */
+void feat_vec_fill_input_array_dnn(float *input_array, feat_vec *fv, feat_model *fm, mcd *m)
+{
+  int i,j;
+  feat_desc *fd;
+  simple_feat_desc *sfd;
+  int mcd_representation;
+  int current_index = 0;
+  int column;
+
+  for(i=0; i < fm->nbelem; i++){
+    fd = fm->array[i];
+    sfd = fd->array[0];
+
+    if((sfd->type <= FEAT_TYPE_INT_16) && (sfd->type >= FEAT_TYPE_INT_1)){
+      int range = sfd->type - FEAT_TYPE_INT_0;
+      for(j=0; j < range; j++){
+	if(j == fv->t[i])
+	  input_array[current_index++] = 1;
+	else
+	  input_array[current_index++] = 0;
+      }
+    }
+    else{
+      column = m->wf2col[sfd->type];
+      mcd_representation = m->representation[column];
+      if(mcd_representation == MCD_REPRESENTATION_EMB){
+	current_index =  word_emb_fill_input_array_dnn(input_array, m->word_emb_array[column], fv->t[i], current_index);
+      }
+      else if(mcd_representation == MCD_REPRESENTATION_VOCAB){
+	for(j=0; j < m->dico_array[column]->nbelem; j++){
+	  if(j == fv->t[i])
+	    input_array[current_index++] = 1;
+	  else
+	    input_array[current_index++] = 0;
+	}
+      }
+    }
+ 
+  }
+}
+
+

perceptron/CMakeLists.txt

+add_subdirectory(lib)
+add_subdirectory(exec)

perceptron/exec/CMakeLists.txt

+#compiling, linking and installing executables
+
+add_executable(perceptron_train perceptron_train.c)
+target_link_libraries(perceptron_train perceptron)
+target_link_libraries(perceptron_train maca_common)
+install (TARGETS perceptron_train DESTINATION bin)
+
+#add_executable(maca_trans_parser_cff_cutoff cff_cutoff.c)
+#target_link_libraries(maca_trans_parser_cff_cutoff perceptron)
+#target_link_libraries(maca_trans_parser_cff_cutoff maca_common)
+#install (TARGETS maca_trans_parser_cff_cutoff DESTINATION bin)
+
+add_executable(perceptron_eval perceptron_eval.c)
+target_link_libraries(perceptron_eval perceptron)
+target_link_libraries(perceptron_eval maca_common)
+install (TARGETS perceptron_eval DESTINATION bin)

maca_trans_parser/src/cff_cutoff.c → perceptron/exec/cff_cutoff.c

@@ -3,21 +3,18 @@
 #include<string.h>
 #include<unistd.h>
 #include<getopt.h>
-#include"movement_parser.h"
-#include"oracle_parser.h"
 #include"feature_table.h"
 #include"dico.h"
 #include"perceptron.h"
-#include"context.h"
+#include"perceptron_context.h"
 #include"cf_file.h"
 
-void cff_cutoff_help_message(context *ctx)
+void cff_cutoff_help_message(perceptron_context *ctx)
 {
-  context_general_help_message(ctx);
+  perceptron_context_help_message(ctx);
   fprintf(stderr, "INPUT\n");
-  context_cutoff_help_message(ctx);
-  context_cff_help_message(ctx);
-  fprintf(stderr, "INPUT/OUTPUT\n");
+  perceptron_context_cutoff_help_message(ctx);
+
 }
 
 void cff_cutoff_check_options(context *ctx)

maca_trans_parser/src/eval_classifier.c → perceptron/exec/perceptron_eval.c

@@ -3,19 +3,20 @@
 #include<string.h>
 #include<unistd.h>
 #include<getopt.h>
+#include"util.h"
 #include"feature_table.h"
 #include"perceptron.h"
-#include"context.h"
+#include"perceptron_context.h"
 
-void eval_classifier_help_message(context *ctx)
+void eval_classifier_help_message(perceptron_context *ctx)
 {
-  context_general_help_message(ctx);
-  fprintf(stderr, "INPUT\n");
-  context_model_help_message(ctx);
-  context_cff_help_message(ctx);
+  perceptron_context_help_message(ctx);
+  /* fprintf(stderr, "INPUT\n"); */
+  /* perceptron_context_model_help_message(ctx); */
+  /* perceptron_context_cff_help_message(ctx); */
 }
 
-void eval_classifier_check_options(context *ctx)
+void eval_classifier_check_options(perceptron_context *ctx)
 {
   if(!ctx->cff_filename
      || !ctx->perc_model_filename
@@ -37,10 +38,10 @@ int main(int argc, char *argv[])
   feat_vec *fv = feat_vec_new(2);
   char *token;
   float accuracy;
-  context *ctx;
+  perceptron_context *ctx;
   float score;
 
-  ctx = context_read_options(argc, argv);
+  ctx = perceptron_context_read_options(argc, argv);
   eval_classifier_check_options(ctx);
 
   ft = feature_table_load(ctx->perc_model_filename, ctx->verbose);

maca_trans_parser/src/train.c → perceptron/exec/perceptron_train.c

@@ -3,26 +3,18 @@
 #include<string.h>
 #include<unistd.h>
 #include<getopt.h>
-#include"movement_parser.h"
-#include"oracle_parser.h"
-#include"feat_fct.h"
 #include"feature_table.h"
 #include"dico.h"
 #include"perceptron.h"
-#include"context.h"
+#include"perceptron_context.h"
 #include"cf_file.h"
 
-void train_help_message(context *ctx)
+void train_help_message(perceptron_context *ctx)
 {
-  context_general_help_message(ctx);
-  context_iterations_help_message(ctx);
-  fprintf(stderr, "INPUT\n");
-  context_cff_help_message(ctx);
-  fprintf(stderr, "OUTPUT\n");
-  context_model_help_message(ctx);
+  perceptron_context_help_message(ctx);
 }
 
-void train_check_options(context *ctx)
+void train_check_options(perceptron_context *ctx)
 {
   if(!ctx->cff_filename
      || !ctx->perc_model_filename
@@ -37,9 +29,9 @@ int main(int argc, char *argv[])
 {
   int nb_feat, nb_class;
   feature_table *ft;
-  context *ctx;
+  perceptron_context *ctx;
 
-  ctx = context_read_options(argc, argv);
+  ctx = perceptron_context_read_options(argc, argv);
   train_check_options(ctx);
 
   look_for_number_of_features_and_classes(ctx->cff_filename, &nb_feat, &nb_class);
@@ -49,7 +41,7 @@ int main(int argc, char *argv[])
   perceptron_avg(ctx->cff_filename, ft, ctx->iteration_nb);
   feature_table_dump(ctx->perc_model_filename, ft);
   
-  context_free(ctx);
+  perceptron_context_free(ctx);
 
   return 0;
 }

perceptron/lib/CMakeLists.txt

+set(SOURCES  src/cf_file.c
+  src/feature_table.c
+  src/perceptron.c
+  src/feat_vec.c
+  src/perceptron_context.c
+)
+
+#compiling library
+include_directories(src)
+add_library(perceptron STATIC ${SOURCES})

perceptron/lib/include/perceptron_context.h

+#ifndef __PERCEPTRON_CONTEXT__
+#define __PERCEPTRON_CONTEXT__
+
+#include "dico_vec.h"
+#include<stdlib.h>
+
+typedef struct {
+  int help;
+  int verbose;
+  char *program_name;
+  char *cff_filename;
+  char *perc_model_filename;
+  int iteration_nb;
+} perceptron_context;
+
+perceptron_context *perceptron_context_new(void);
+void perceptron_context_free(perceptron_context *ctx);
+perceptron_context *perceptron_context_read_options(int argc, char *argv[]);
+void perceptron_context_help_message(perceptron_context *ctx);
+
+
+#endif

maca_trans_parser/src/feat_vec.c → perceptron/lib/src/feat_vec.c

@@ -2,8 +2,8 @@
 #include<stdlib.h>
 #include"feat_vec.h"
 #include"util.h"
-#include "feat_types.h"
-#include "feat_model.h"
+/* #include "feat_types.h" */
+/* #include "feat_model.h" */
 
 void feat_vec_concat(feat_vec *fv1, feat_vec *fv2)
 {
@@ -78,90 +78,3 @@ void feat_vec_print(FILE *f, feat_vec *fv)
   fprintf(f, "\n");
 }
 
-void feat_vec_print_dnn(FILE *f, feat_vec *fv, feat_model *fm, mcd *m)
-{
-  int i,j;
-  feat_desc *fd;
-  simple_feat_desc *sfd;
-  int mcd_representation;
-  int column;
-
-  for(i=0; i < fm->nbelem; i++){
-    fd = fm->array[i];
-    sfd = fd->array[0];
-    
-    if((sfd->type <= FEAT_TYPE_INT_16) && (sfd->type >= FEAT_TYPE_INT_1)){
-      int range = sfd->type - FEAT_TYPE_INT_0;
-      for(j=0; j < range; j++){
-	/* if(i + j != 0) fprintf(f, " "); */
-	if(j == fv->t[i])
-	  fprintf(f, "1 ");
-	else
-	  fprintf(f, "0 ");
-      }
-    }
-    else{
-      column = m->wf2col[sfd->type];
-      mcd_representation = m->representation[column];
-      if(mcd_representation == MCD_REPRESENTATION_EMB){
-	/* if(i != 0) fprintf(f, " "); */
-	word_emb_print(f, m->word_emb_array[column], fv->t[i]);
-      }
-      else if(mcd_representation == MCD_REPRESENTATION_VOCAB){
-	for(j=0; j < m->dico_array[column]->nbelem; j++){
-	  /* if(i + j != 0) fprintf(f, " "); */
-	  if(j == fv->t[i])
-	    fprintf(f, "1 ");
-	  else
-	    fprintf(f, "0 ");
-	}
-      }
-    }
-  fprintf(f, "\n");
-
-  }
-  fprintf(f, "\n");
-}
-
-/* void feat_vec_fill_input_array_dnn(fann_type *input_array, feat_vec *fv, feat_model *fm, mcd *m) */
-void feat_vec_fill_input_array_dnn(float *input_array, feat_vec *fv, feat_model *fm, mcd *m)
-{
-  int i,j;
-  feat_desc *fd;
-  simple_feat_desc *sfd;
-  int mcd_representation;
-  int current_index = 0;
-  int column;
-
-  for(i=0; i < fm->nbelem; i++){
-    fd = fm->array[i];
-    sfd = fd->array[0];
-
-    if((sfd->type <= FEAT_TYPE_INT_16) && (sfd->type >= FEAT_TYPE_INT_1)){
-      int range = sfd->type - FEAT_TYPE_INT_0;
-      for(j=0; j < range; j++){
-	if(j == fv->t[i])
-	  input_array[current_index++] = 1;
-	else
-	  input_array[current_index++] = 0;
-      }
-    }
-    else{
-      column = m->wf2col[sfd->type];
-      mcd_representation = m->representation[column];
-      if(mcd_representation == MCD_REPRESENTATION_EMB){
-	current_index =  word_emb_fill_input_array_dnn(input_array, m->word_emb_array[column], fv->t[i], current_index);
-      }
-      else if(mcd_representation == MCD_REPRESENTATION_VOCAB){
-	for(j=0; j < m->dico_array[column]->nbelem; j++){
-	  if(j == fv->t[i])
-	    input_array[current_index++] = 1;
-	  else
-	    input_array[current_index++] = 0;
-	}
-      }
-    }
- 
-  }
-}
-

perceptron/lib/src/perceptron_context.c

+#include<stdlib.h>
+#include<stdio.h>
+#include<string.h>
+#include<unistd.h>
+#include<getopt.h>
+#include "perceptron_context.h"
+#include "util.h"
+
+void perceptron_context_free(perceptron_context *ctx)
+{
+  if(ctx->program_name)            free(ctx->program_name);
+  if(ctx->cff_filename)            free(ctx->cff_filename);
+  if(ctx->perc_model_filename)     free(ctx->perc_model_filename);
+
+  free(ctx);
+}
+
+perceptron_context *perceptron_context_new(void)
+{
+  perceptron_context *ctx = (perceptron_context *)memalloc(sizeof(perceptron_context));
+
+  ctx->verbose = 0;
+  ctx->program_name = NULL;
+  ctx->perc_model_filename = NULL;
+  ctx->cff_filename = NULL;
+  ctx->iteration_nb = 4;
+
+  return ctx;
+}
+
+void perceptron_context_help_message(perceptron_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-n --iter           <int> : number of iterations  (default is 4)\n");
+    fprintf(stderr, "\t-c --cff           <file> : CFF format file name\n");
+    fprintf(stderr, "\t-m --model         <file> : model file name\n");
+}
+
+perceptron_context *perceptron_context_read_options(int argc, char *argv[])
+{
+  int c;
+  int option_index = 0;
+  perceptron_context *ctx = perceptron_context_new();
+
+  ctx->program_name = strdup(argv[0]);
+
+  static struct option long_options[5] =
+    {
+      {"help",                no_argument,       0, 'h'},
+      {"verbose",             no_argument,       0, 'v'},
+      {"model",               required_argument, 0, 'm'}, 
+      {"iter",                required_argument, 0, 'n'},
+      {"cff",                 required_argument, 0, 'c'}
+    };
+  optind = 0;
+  opterr = 0;
+  
+  
+  while ((c = getopt_long (argc, argv, "hvm:n:c:", long_options, &option_index)) != -1){ 
+    switch (c)
+      {
+      case 'h':
+	ctx->help = 1;
+	break;
+      case 'v':
+	ctx->verbose = 1;
+	break;
+      case 'm':
+	ctx->perc_model_filename = strdup(optarg);
+	break;
+      case 'n':
+	ctx->iteration_nb = atoi(optarg);
+	break;
+      case 'c':
+	ctx->cff_filename = strdup(optarg);
+	break;
+      }
+  }
+
+  return ctx;
+}
+
+