From ebabce6be0cb28ffd0cdf50093c4b4d3d7cdae8a Mon Sep 17 00:00:00 2001
From: Alexis Nasr <alexis.nasr@lif.univ-mrs.fr>
Date: Tue, 6 Dec 2016 10:10:07 -0500
Subject: [PATCH] added maca_trans_tagger_bt an (experimental) tagger with a
(very) limited for of backtrack, modfied some help messages
---
maca_common/include/word_buffer.h | 5 -
maca_trans_parser/src/context.c | 3 +
maca_trans_parser/src/context.h | 3 +
maca_trans_parser/src/maca_trans_parser.c | 9 +-
maca_trans_parser/src/maca_trans_tagger_bt.c | 103 +++++++++++++++++++
perceptron/lib/include/feat_vec.h | 14 ++-
perceptron/lib/include/feature_table.h | 20 ++--
perceptron/lib/src/feature_table.c | 2 -
8 files changed, 130 insertions(+), 29 deletions(-)
create mode 100644 maca_trans_parser/src/maca_trans_tagger_bt.c
diff --git a/maca_common/include/word_buffer.h b/maca_common/include/word_buffer.h
index 91b216d..0c30a9a 100644
--- a/maca_common/include/word_buffer.h
+++ b/maca_common/include/word_buffer.h
@@ -29,11 +29,6 @@
#define word_buffer_is_last(wb) (((wb)->current_index == (wb)->nbelem - 1)? 1 : 0)
#define word_buffer_is_empty(wb) (((wb)->nbelem == 0)? 1 : 0)
-
-
-
-
-
typedef struct {
int size; /* size of the array used to store words */
int nbelem; /* number of words in the buffer */
diff --git a/maca_trans_parser/src/context.c b/maca_trans_parser/src/context.c
index 85acc24..72c2d61 100644
--- a/maca_trans_parser/src/context.c
+++ b/maca_trans_parser/src/context.c
@@ -164,6 +164,9 @@ void context_f2p_filename_help_message(context *ctx){
void context_trace_mode_help_message(context *ctx){
fprintf(stderr, "\t-T --traces : activate trace mode (default is false)\n");
}
+void context_debug_help_message(context *ctx){
+ fprintf(stderr, "\t-d --debug : activate debug mode (default is false)\n");
+}
context *context_read_options(int argc, char *argv[])
{
diff --git a/maca_trans_parser/src/context.h b/maca_trans_parser/src/context.h
index 611dd10..932e671 100644
--- a/maca_trans_parser/src/context.h
+++ b/maca_trans_parser/src/context.h
@@ -102,4 +102,7 @@ void context_ifpls_help_message(context *ctx);
void context_input_help_message(context *ctx);
void context_root_label_help_message(context *ctx);
+void context_debug_help_message(context *ctx);
+
+
#endif
diff --git a/maca_trans_parser/src/maca_trans_parser.c b/maca_trans_parser/src/maca_trans_parser.c
index 3ecbcd4..fa44e9b 100644
--- a/maca_trans_parser/src/maca_trans_parser.c
+++ b/maca_trans_parser/src/maca_trans_parser.c
@@ -13,11 +13,12 @@
/*#include"dnn_decoder.h"*/
#include"config2feat_vec.h"
-void decode_help_message(context *ctx)
+void maca_trans_parser_help_message(context *ctx)
{
context_general_help_message(ctx);
/* context_beam_help_message(ctx); */
/* context_conll_help_message(ctx); */
+ context_debug_help_message(ctx);
fprintf(stderr, "INPUT\n");
context_input_help_message(ctx);
context_mcd_help_message(ctx);
@@ -27,7 +28,7 @@ void decode_help_message(context *ctx)
context_root_label_help_message(ctx);
}
-void decode_check_options(context *ctx){
+void maca_trans_parser_check_options(context *ctx){
if(ctx->help
/*!ctx->conll_filename*/
/* || !ctx->perc_model_filename
@@ -35,7 +36,7 @@ void decode_check_options(context *ctx){
|| !ctx->vocabs_filename
|| !ctx->features_model_filename*/
){
- decode_help_message(ctx);
+ maca_trans_parser_help_message(ctx);
exit(1);
}
}
@@ -83,7 +84,7 @@ int main(int argc, char *argv[])
context *ctx;
ctx = context_read_options(argc, argv);
- decode_check_options(ctx);
+ maca_trans_parser_check_options(ctx);
set_linguistic_resources_filenames_parser(ctx);
ctx->features_model = feat_model_read(ctx->features_model_filename, ctx->verbose);
diff --git a/maca_trans_parser/src/maca_trans_tagger_bt.c b/maca_trans_parser/src/maca_trans_tagger_bt.c
new file mode 100644
index 0000000..778c634
--- /dev/null
+++ b/maca_trans_parser/src/maca_trans_tagger_bt.c
@@ -0,0 +1,103 @@
+#include<stdio.h>
+#include<stdlib.h>
+#include<string.h>
+#include<unistd.h>
+#include<getopt.h>
+#include"context.h"
+#include"feat_fct.h"
+#include"feature_table.h"
+#include"dico.h"
+#include"beam.h"
+#include"form2pos.h"
+#include"simple_decoder_tagger.h"
+/*#include"dnn_decoder.h"*/
+#include"config2feat_vec.h"
+
+void decode_tagger_help_message(context *ctx)
+{
+ context_general_help_message(ctx);
+ context_beam_help_message(ctx);
+ context_conll_help_message(ctx);
+ fprintf(stderr, "INPUT\n");
+ context_input_help_message(ctx);
+ context_mcd_help_message(ctx);
+ context_model_help_message(ctx);
+ context_vocabs_help_message(ctx);
+ context_features_model_help_message(ctx);
+ context_f2p_filename_help_message(ctx);
+}
+
+void decode_tagger_check_options(context *ctx){
+ if(ctx->help
+ /*!ctx->conll_filename*/
+ /* || !ctx->perc_model_filename
+ || !ctx->mcd_filename
+ || !ctx->vocabs_filename
+ || !ctx->features_model_filename*/
+ ){
+ decode_tagger_help_message(ctx);
+ exit(1);
+ }
+}
+
+void decode_tagger_set_linguistic_resources_filenames(context *ctx)
+{
+ char absolute_filename[500];
+
+ if(!ctx->perc_model_filename){
+ strcpy(absolute_filename, ctx->maca_data_path);
+ strcat(absolute_filename, DEFAULT_MODEL_TAGGER_FILENAME);
+ ctx->perc_model_filename = strdup(absolute_filename);
+ }
+
+ if(!ctx->vocabs_filename){
+ strcpy(absolute_filename, ctx->maca_data_path);
+ strcat(absolute_filename, DEFAULT_VOCABS_TAGGER_FILENAME);
+ ctx->vocabs_filename = strdup(absolute_filename);
+ }
+
+ /* if(!ctx->mcd_filename){
+ strcpy(absolute_filename, ctx->maca_data_path);
+ strcat(absolute_filename, DEFAULT_MULTI_COL_DESC_TAGGER_FILENAME);
+ ctx->mcd_filename = strdup(absolute_filename);
+ }*/
+
+ if(!ctx->features_model_filename){
+ strcpy(absolute_filename, ctx->maca_data_path);
+ strcat(absolute_filename, DEFAULT_FEATURES_MODEL_TAGGER_FILENAME);
+ ctx->features_model_filename = strdup(absolute_filename);
+ }
+
+ if(!ctx->f2p_filename){
+ strcpy(absolute_filename, ctx->maca_data_path);
+ strcat(absolute_filename, DEFAULT_F2P_FILENAME);
+ ctx->f2p_filename = strdup(absolute_filename);
+ ctx->f2p = form2pos_read(ctx->f2p_filename);
+ }
+
+ if(ctx->verbose){
+ 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[])
+{
+ context *ctx = context_read_options(argc, argv);
+ decode_tagger_check_options(ctx);
+
+ decode_tagger_set_linguistic_resources_filenames(ctx);
+ ctx->features_model = feat_model_read(ctx->features_model_filename, ctx->verbose);
+ ctx->vocabs = dico_vec_read(ctx->vocabs_filename, ctx->hash_ratio);
+ mcd_link_to_dico(ctx->mcd_struct, ctx->vocabs, ctx->verbose);
+
+ simple_decoder_tagger2(ctx);
+
+ context_free(ctx);
+ return 0;
+}
+
diff --git a/perceptron/lib/include/feat_vec.h b/perceptron/lib/include/feat_vec.h
index 4116d76..e1dfe91 100644
--- a/perceptron/lib/include/feat_vec.h
+++ b/perceptron/lib/include/feat_vec.h
@@ -13,16 +13,14 @@ typedef struct {
/*#include "word_emb.h"*/
#include "mcd.h"
-
-void feat_vec_concat(feat_vec *fv1, feat_vec *fv2);
+void feat_vec_concat(feat_vec *fv1, feat_vec *fv2);
feat_vec *feat_vec_copy(feat_vec *fv);
-
feat_vec *feat_vec_new(int size);
-void feat_vec_free(feat_vec *fv);
-int feat_vec_add(feat_vec *fv, int feat);
-void feat_vec_empty(feat_vec *fv);
-void feat_vec_print_string(feat_vec *fv, dico *dico_features);
-void feat_vec_print(FILE *f, feat_vec *fv);
+void feat_vec_free(feat_vec *fv);
+int feat_vec_add(feat_vec *fv, int feat);
+void feat_vec_empty(feat_vec *fv);
+void feat_vec_print_string(feat_vec *fv, dico *dico_features);
+void feat_vec_print(FILE *f, feat_vec *fv);
/* void feat_vec_print_dnn(FILE *f, feat_vec *fv, feat_model *fm, mcd *m); */
/* 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); */
diff --git a/perceptron/lib/include/feature_table.h b/perceptron/lib/include/feature_table.h
index ef1bbe6..97abb90 100644
--- a/perceptron/lib/include/feature_table.h
+++ b/perceptron/lib/include/feature_table.h
@@ -17,15 +17,15 @@ typedef struct {
} vcode;
feature_table *feature_table_load(char *filename, int verbose);
-void feature_table_dump(char *filename, feature_table *ft);
+void feature_table_dump(char *filename, feature_table *ft);
feature_table *feature_table_new(int features_nb, int classes_nb);
-void feature_table_print(char *filename, feature_table *ft);
-void feature_table_print_verbose(char *filename, feature_table *ft, dico *dico_features, dico *dico_classes);
-int feature_table_argmax(feat_vec *fv, feature_table *ft, float *max);
-float feature_table_entropy(feat_vec *fv, feature_table *ft);
-float feature_table_diff_scores(feat_vec *fv, feature_table *ft);
-float feature_table_argmax_1_2(feat_vec *fv, feature_table *ft, int *argmax1, float *max1, int *argmax2, float *max2);
-void feature_table_free(feature_table *ft);
-void feature_table_scores(feat_vec *fv, feature_table *ft, float *classes_score);
-vcode* feature_table_get_vcode_array(feat_vec *fv, feature_table* ft);
+void feature_table_print(char *filename, feature_table *ft);
+void feature_table_print_verbose(char *filename, feature_table *ft, dico *dico_features, dico *dico_classes);
+int feature_table_argmax(feat_vec *fv, feature_table *ft, float *max);
+float feature_table_entropy(feat_vec *fv, feature_table *ft);
+float feature_table_diff_scores(feat_vec *fv, feature_table *ft);
+float feature_table_argmax_1_2(feat_vec *fv, feature_table *ft, int *argmax1, float *max1, int *argmax2, float *max2);
+void feature_table_free(feature_table *ft);
+void feature_table_scores(feat_vec *fv, feature_table *ft, float *classes_score);
+vcode *feature_table_get_vcode_array(feat_vec *fv, feature_table* ft);
#endif
diff --git a/perceptron/lib/src/feature_table.c b/perceptron/lib/src/feature_table.c
index 5c2384a..248a1b3 100644
--- a/perceptron/lib/src/feature_table.c
+++ b/perceptron/lib/src/feature_table.c
@@ -8,8 +8,6 @@
feature_table *feature_table_load(char *filename, int verbose)
{
int i;
-
-
feature_table *ft = NULL;
int features_nb;
int classes_nb;
--
GitLab