diff --git a/maca_common/include/word.h b/maca_common/include/word.h
index 7620b6d8263d46fb1e240f9a72998c9521e7ba8e..3990714bf0e6ec67c62dbbc1a5a7fe83a25e23e3 100644
--- a/maca_common/include/word.h
+++ b/maca_common/include/word.h
@@ -5,7 +5,7 @@
#define WORD_INVALID_GOV 10000
-#define word_get_id(w) ((w) == NULL) ? -1 : (w)->wf_array[MCD_WF_ID]
+#define word_get_id(w) ((w) == NULL) ? -1 : (w)->wf_array[MCD_WF_ID]
#define word_get_form(w) ((w) == NULL) ? -1 : (w)->wf_array[MCD_WF_FORM]
#define word_get_lemma(w) ((w) == NULL) ? -1 : (w)->wf_array[MCD_WF_LEMMA]
#define word_get_cpos(w) ((w) == NULL) ? -1 : (w)->wf_array[MCD_WF_CPOS]
diff --git a/maca_common/include/word_buffer.h b/maca_common/include/word_buffer.h
index 2b182b73f6d0b91e857b5035764ee8258f27750f..bce1b7d90fdc33d21ff4395834e2db96620e1087 100644
--- a/maca_common/include/word_buffer.h
+++ b/maca_common/include/word_buffer.h
@@ -13,15 +13,15 @@
#define word_buffer_get_input_file(wb) (wb)->input_file
#define word_buffer_get_mcd(wb) (wb)->mcd_struct
-#define word_buffer_b0(wb) ((wb)->nbelem == 0)? NULL : (wb)->array[(wb)->current_index]
-#define word_buffer_b1(wb) ((wb)->current_index + 1 >= (wb)->nbelem)? NULL : (wb)->array[(wb)->current_index + 1]
-#define word_buffer_b2(wb) ((wb)->current_index + 2 >= (wb)->nbelem)? NULL : (wb)->array[(wb)->current_index + 2]
-#define word_buffer_b3(wb) ((wb)->current_index + 3 >= (wb)->nbelem)? NULL : (wb)->array[(wb)->current_index + 3]
-#define word_buffer_bm1(wb) ((wb)->current_index - 1 < 0)? NULL : (wb)->array[(wb)->current_index - 1]
-#define word_buffer_bm2(wb) ((wb)->current_index - 2 < 0)? NULL : (wb)->array[(wb)->current_index - 2]
-#define word_buffer_bm3(wb) ((wb)->current_index - 3 < 0)? NULL : (wb)->array[(wb)->current_index - 3]
-#define word_buffer_nb_elts_right(wb) (wb)->nbelem - (wb)->current_index - 1
-#define word_buffer_nb_elts_left(wb) (wb)->current_index
+#define word_buffer_b0(wb) (((wb)->nbelem == 0)? NULL : (wb)->array[(wb)->current_index])
+#define word_buffer_b1(wb) (((wb)->current_index + 1 >= (wb)->nbelem)? NULL : (wb)->array[(wb)->current_index + 1])
+#define word_buffer_b2(wb) (((wb)->current_index + 2 >= (wb)->nbelem)? NULL : (wb)->array[(wb)->current_index + 2])
+#define word_buffer_b3(wb) (((wb)->current_index + 3 >= (wb)->nbelem)? NULL : (wb)->array[(wb)->current_index + 3])
+#define word_buffer_bm1(wb) (((wb)->current_index - 1 < 0)? NULL : (wb)->array[(wb)->current_index - 1])
+#define word_buffer_bm2(wb) (((wb)->current_index - 2 < 0)? NULL : (wb)->array[(wb)->current_index - 2])
+#define word_buffer_bm3(wb) (((wb)->current_index - 3 < 0)? NULL : (wb)->array[(wb)->current_index - 3])
+#define word_buffer_nb_elts_right(wb) ((wb)->nbelem - (wb)->current_index - 1)
+#define word_buffer_nb_elts_left(wb) ((wb)->current_index)
typedef struct {
diff --git a/maca_trans_parser/CMakeLists.txt b/maca_trans_parser/CMakeLists.txt
index 6ad73f7fd8743d5a5e9e0ce18ccce04e5ad4856f..85d67be561ad16cd6349bf20b8cbec8c62aadf57 100644
--- a/maca_trans_parser/CMakeLists.txt
+++ b/maca_trans_parser/CMakeLists.txt
@@ -1,31 +1,32 @@
set(SOURCES src/context.c
- src/feat_desc.c
-src/movement_parser_arc_eager.c
- src/movement_tagparser_arc_eager.c
- src/movement_tagger.c
- src/feat_fct.c
- src/global_feat_vec.c
-# src/oracle_parser.c
- src/oracle_parser_arc_eager.c
- src/oracle_tagparser_arc_eager.c
- src/oracle_tagger.c
- src/simple_decoder_parser.c
- src/simple_decoder_parser_arc_eager.c
- src/simple_decoder_tagparser_arc_eager.c
- src/simple_decoder_forrest.c
- src/simple_decoder_tagger.c
- src/feat_lib.c
- src/stack.c
- src/config2feat_vec.c
- src/depset.c
- src/feat_model.c
- src/config.c
- src/queue.c
-# src/beam.c
-src/feat_types.c
-src/mvt.c
-src/mvt_stack.c
-)
+ src/feat_desc.c
+ src/movements.c
+ src/movement_parser_arc_eager.c
+ src/movement_tagparser_arc_eager.c
+ src/movement_tagger.c
+ src/feat_fct.c
+ src/global_feat_vec.c
+ # src/oracle_parser.c
+ src/oracle_parser_arc_eager.c
+ src/oracle_tagparser_arc_eager.c
+ src/oracle_tagger.c
+# src/simple_decoder_parser.c
+ src/simple_decoder_parser_arc_eager.c
+ src/simple_decoder_tagparser_arc_eager.c
+# src/simple_decoder_forrest.c
+ src/simple_decoder_tagger.c
+ src/feat_lib.c
+ src/stack.c
+ src/config2feat_vec.c
+ src/depset.c
+ src/feat_model.c
+ src/config.c
+# src/queue.c
+ # src/beam.c
+ src/feat_types.c
+ src/mvt.c
+ src/mvt_stack.c
+ )
#compiling library
include_directories(src)
@@ -82,11 +83,11 @@ target_link_libraries(maca_trans_tagparser transparse)
target_link_libraries(maca_trans_tagparser maca_common)
install (TARGETS maca_trans_tagparser 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_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)
diff --git a/maca_trans_parser/src/cff2fann.c b/maca_trans_parser/src/cff2fann.c
new file mode 100644
index 0000000000000000000000000000000000000000..2167fa8d0108633ea55e3b3802bb271d3b94157d
--- /dev/null
+++ b/maca_trans_parser/src/cff2fann.c
@@ -0,0 +1,113 @@
+#include<stdio.h>
+#include<stdlib.h>
+#include<string.h>
+#include<unistd.h>
+#include<getopt.h>
+#include"context.h"
+#include"util.h"
+#include"cf_file.h"
+
+
+void cff2fann_help_message(context *ctx)
+{
+ context_general_help_message(ctx);
+ context_mode_help_message(ctx);
+ context_sent_nb_help_message(ctx);
+
+ fprintf(stderr, "INPUT\n");
+ context_conll_help_message(ctx);
+ fprintf(stderr, "IN TEST MODE\n");
+ context_vocabs_help_message(ctx);
+
+ fprintf(stderr, "OUTPUT\n");
+ context_cff_help_message(ctx);
+ fprintf(stderr, "IN TRAIN MODE\n");
+ context_vocabs_help_message(ctx);
+
+}
+
+void cff2fann_check_options(context *ctx)
+{
+ if(!ctx->input_filename
+ || ctx->help
+ /* || !ctx->mcd_filename */
+ /* || !(ctx->cff_filename || ctx->fann_filename) */
+ ){
+ cff2fann_help_message(ctx);
+ exit(1);
+ }
+}
+
+void one_hot_print(FILE *f, int val, int dim)
+{
+ int i;
+ for(i=0; i < dim; i++)
+ fprintf(f, "%d ", (i == val)? 1 : 0);
+}
+
+void cff2fann(context *ctx)
+{
+ char buffer[10000];
+ char *token;
+ int col_nb;
+ int feat_type;
+ mcd *m = ctx->mcd_struct;
+ FILE *f = myfopen(ctx->input_filename, "r");
+ int val;
+
+ while(fgets(buffer, 10000, f)){
+ /* printf("%s", buffer); */
+ /* printf("\n"); */
+ token = strtok(buffer, "\t");
+ col_nb = 0;
+ while(token){
+ /* printf("col = %d token = %s max = %d\n", col_nb, token, max_array[col_nb]); */
+ val = atoi(token);
+ if(col_nb == 0){
+ one_hot_print(stdout, val, ctx->mvt_nb);
+ printf("\n");
+ }
+ else{
+ feat_type = feat_model_get_type_feat_n(ctx->features_model, col_nb - 1);
+ /* printf("feat_type = %d\n", feat_type); */
+ int mcd_col = m->wf2col[feat_type];
+ /* printf("representation = %d\n", m->representation[mcd_col]); */
+ if(m->representation[mcd_col] == MCD_REPRESENTATION_EMB){
+ /* printf("it is an embedding val = %d, file = %s\n", val, m->filename[mcd_col]); */
+ word_emb_print(stdout, m->word_emb_array[mcd_col], val);
+ printf("\n");
+ }
+ if(m->representation[mcd_col] == MCD_REPRESENTATION_VOCAB){
+ /* printf("it is a vocab\n"); */
+ one_hot_print(stdout, val, m->dico_array[mcd_col]->nbelem);
+ printf("\n");
+ }
+ }
+ col_nb++;
+ token = strtok(NULL , "\t");
+ }
+ }
+}
+
+int main(int argc, char *argv[])
+{
+ context *ctx;
+ int nb_feat;
+ int nb_class;
+
+ ctx = context_read_options(argc, argv);
+ cff2fann_check_options(ctx);
+
+ ctx->vocabs = dico_vec_read(ctx->vocabs_filename, ctx->hash_ratio);
+
+ ctx->features_model = feat_model_read(ctx->features_model_filename, ctx->verbose);
+
+ look_for_number_of_features_and_classes(ctx->input_filename, &nb_feat, &nb_class);
+ ctx->mvt_nb = nb_class;
+
+ mcd_link_to_dico(ctx->mcd_struct, ctx->vocabs, 1);
+
+ cff2fann(ctx);
+ return 0;
+}
+
diff --git a/maca_trans_parser/src/compare_traces.c b/maca_trans_parser/src/compare_traces.c
index a4d1875da2223b891457657293752502712b4f31..f6567c3df39b0938ad658283241576387d3914b7 100644
--- a/maca_trans_parser/src/compare_traces.c
+++ b/maca_trans_parser/src/compare_traces.c
@@ -112,7 +112,7 @@ void trace_compare(trace *ref, trace *hyp)
int index_hyp = 0;
int index_ref = 0;
configuration *c_ref, *c_hyp;
- int status = 1;
+
while(1){
c_ref = ref->array[index_ref];
c_hyp = hyp->array[index_hyp];
@@ -123,7 +123,6 @@ void trace_compare(trace *ref, trace *hyp)
configuration_print(stdout, c_hyp);
if(configuration_equal(c_ref, c_hyp)){
- status = 1;
fprintf(stdout, "EQUAL\n");
if(strcmp(c_ref->movement, c_hyp->movement)){
/* fprintf(stdout, "BAAD\t%s\t%s\t%f\n", c_ref->movement, c_hyp->movement, c_hyp->score); */
@@ -138,7 +137,6 @@ void trace_compare(trace *ref, trace *hyp)
}
else{
fprintf(stdout, "DIFFERENT\n");
- status = 0;
if(c_ref->index > c_hyp->index)
index_hyp++;
else if(c_ref->index < c_hyp->index)
diff --git a/maca_trans_parser/src/config.c b/maca_trans_parser/src/config.c
index 392299edc8070cd6b03f45028154fd4ca7b00ff6..ed5e3f0b7c04f2c636f82573d79ea66389372488 100644
--- a/maca_trans_parser/src/config.c
+++ b/maca_trans_parser/src/config.c
@@ -8,105 +8,35 @@
config *config_new(FILE *f, mcd *mcd_struct, int lookahead)
{
config *c = (config *)memalloc(sizeof(config));
- c->ds = depset_new();
- c->f = f;
- c->mcd_struct = mcd_struct;
c->st = stack_new();
c->bf = word_buffer_new(f, mcd_struct, lookahead);
- c->score = 0;
- c->mvt_array = NULL;
- c->mvt_array_size = 0;
- c->mvt_array_nbelem = 0;
- c->gfv = global_feat_vec_new();
+ c->history = mvt_stack_new();
return c;
}
-/*word *config_add_next_word_to_buffer(config *c)
-{
- word *w = NULL;
-
- w = word_read(c->f, c->mcd_struct);
- if(w == NULL) return NULL;
-
- word_set_index(w, c->current_index);
- c->current_index++;
- queue_add(c->bf, w);
- return w;
- }*/
-
void config_free(config *c)
{
stack_free(c->st);
word_buffer_free(c->bf);
- /* depset_free(c->ds); */
- if(c->mvt_array)free(c->mvt_array);
- if(c->gfv)global_feat_vec_free(c->gfv);
+ mvt_stack_free(c->history);
free(c);
}
-
-
-
-
int config_is_terminal(config *c)
{
- /* return (word_buffer_end(config_get_buffer(c)) && stack_is_empty(config_get_stack(c))); */
-
return (word_buffer_end(config_get_buffer(c)) &&
(stack_is_empty(config_get_stack(c)) ||
(stack_top(config_get_stack(c)) == NULL)));
-
-
- /* return word_buffer_end(config_get_buffer(c)); */
-}
-
-config *config_initial(FILE *f, mcd *mcd_struct, int lookahead)
-{
- config *c = config_new(f, mcd_struct, lookahead);
-
- /* stack_push(c->st, word_create_dummy(mcd_struct)); */
-
- return c;
-}
-
-config *config_initial_no_dummy_word(FILE *f, mcd *mcd_struct, int lookahead)
-{
- config *c = config_new(f, mcd_struct, lookahead);
-
- return c;
}
-/* must be checked things probably missing */
-
-config *config_copy(config *o)
+void config_push_mvt(config *c, int type, word *gov, word *dep)
{
- int i;
- config *c = (config *)memalloc(sizeof(config));
- c->ds = depset_copy(o->ds);
- c->st = stack_copy(o->st);
- /* c->bf = queue_copy(o->bf); */
- c->score = o->score;
-
- c->f = o->f;
- c->mcd_struct = o->mcd_struct;
-
- c->mvt_array_size = o->mvt_array_size;
- c->mvt_array_nbelem = o->mvt_array_nbelem;
- c->mvt_array = (int *)memalloc(c->mvt_array_size * sizeof(int));
- for(i=0; i < c->mvt_array_nbelem; i++)
- c->mvt_array[i] = o->mvt_array[i];
-
- c->gfv = global_feat_vec_copy(o->gfv);
- return c;
+ mvt_stack_push(config_get_history(c), mvt_new(type, gov, dep));
}
-void config_add_mvt(config *c, int mvt)
+mvt *config_pop_mvt(config *c)
{
- if(c->mvt_array_size == c->mvt_array_nbelem){
- c->mvt_array_size = 2 * (c->mvt_array_size + 1);
- c->mvt_array = (int *)realloc(c->mvt_array, c->mvt_array_size * sizeof(int));
- }
- c->mvt_array[c->mvt_array_nbelem++] = mvt;
+ return mvt_stack_pop(config_get_history(c));
}
void config_print(FILE *f, config *c)
@@ -121,48 +51,3 @@ void config_print(FILE *f, config *c)
}
}
-void config_connect_subtrees(config *c, int root_label)
-{
- int i;
-
- if(stack_height(c->st) > 1){ /* stack contains other elements that dummy token */
- /* make all elements in the stack dependents of the dummy token with a root_label relation */
- for(i=1; i < stack_height(c->st); i++){
- depset_add(c->ds, c->st->array[0], root_label, c->st->array[i]);
- }
- }
-}
-
-int config_equal2(config *c1, config *c2)
-{
- int i;
-
- fprintf(stdout, "pred score = %.4f\n", c1->score);
- fprintf(stdout, "mvt seq ");
- for(i=0; i < c1->mvt_array_nbelem; i++)
- fprintf(stdout, "%d ", c1->mvt_array[i]);
- fprintf(stdout, "\n");
-
- fprintf(stdout, "dep set ");
- depset_print(stdout, c1->ds);
-
- return depset_compare(c1->ds, c2->ds);
-}
-
-int config_equal(config *c1, config *c2)
-{
- int i;
-
- /* for(i=0; i < c2->mvt_array_nbelem; i++)
- fprintf(stdout, "%d ", c2->mvt_array[i]);
- fprintf(stdout, "\n");*/
-
-
- if(c1->mvt_array_nbelem != c2->mvt_array_nbelem)
- return 0;
- for(i=0; i < c1->mvt_array_nbelem; i++){
- if(c1->mvt_array[i] != c2->mvt_array[i])
- return 0;
- }
- return 1;
-}
diff --git a/maca_trans_parser/src/config.h b/maca_trans_parser/src/config.h
index b122662efcc8a5915adf410187fa720c58e68b3a..e2075925265642efe2612b14ec5de129812ced03 100644
--- a/maca_trans_parser/src/config.h
+++ b/maca_trans_parser/src/config.h
@@ -5,45 +5,25 @@
#include<stdlib.h>
#include"stack.h"
-#include"queue.h"
-#include"depset.h"
#include"mcd.h"
#include"word_buffer.h"
-#include"feat_vec.h"
-#include"global_feat_vec.h"
+#include"mvt_stack.h"
#define config_get_stack(c) (c)->st
#define config_get_buffer(c) (c)->bf
-#define config_get_depset(c) (c)->ds
+#define config_get_history(c) (c)->history
typedef struct {
stack *st; /* the stack */
word_buffer *bf; /* the buffer */
- depset *ds; /* the dependency set */
- FILE *f; /* file from which words are read */
- mcd *mcd_struct; /* description of the format of file f */
- float score; /* score of the configuration */
- int *mvt_array; /* array containing the movement sequence that led to this configuration */
- int mvt_array_size; /* self explanatory */
- int mvt_array_nbelem; /* self explanatory */
- global_feat_vec *gfv; /* global feature vector for structured perceptron */
- int current_index; /* index of current word (useful for stream parsing only) */
-
+ mvt_stack *history; /* movement sequence that led to this configuration */
} config;
-int config_equal(config *c1, config *c2);
-int config_equal2(config *c1, config *c2);
config *config_new(FILE *f, mcd *mcd_struct, int lookahead);
-config *config_initial(FILE *f, mcd *mcd_struct, int lookahead);
-config *config_initial_no_dummy_word(FILE *f, mcd *mcd_struct, int lookahead);
-config *config_copy(config *o);
void config_print(FILE *buffer, config *c);
int config_is_terminal(config *c);
void config_free(config *c);
-void config_add_mvt(config *c, int mvt);
-word *config_add_next_word_to_buffer(config *c);
-void config_connect_subtrees(config *c, int root_label);
-
-
+void config_push_mvt(config *c, int type, word *gov, word *dep);
+mvt *config_pop_mvt(config *c);
#endif
diff --git a/maca_trans_parser/src/feat_fct.c b/maca_trans_parser/src/feat_fct.c
index 54cb8ee72da896eab9b28b57ae19bb20a5c581d6..2a24b2e5a90ba3b14315cb2e405f687a4be11c90 100644
--- a/maca_trans_parser/src/feat_fct.c
+++ b/maca_trans_parser/src/feat_fct.c
@@ -597,32 +597,34 @@ int bh(config *c) /* buffer size */
return (config_get_buffer(c)->size > 7)? 7 : config_get_buffer(c)->size;
}
+#if 0
int dh(config *c) /* depset size */
{
return (c->ds->length > 7)? 7 : c->ds->length;
/* return c->ds->length; */
}
+#endif
int t1(config *c) /* previous transition */
{
- if(c->mvt_array_nbelem > 0) return c->mvt_array[c->mvt_array_nbelem - 1];
- return -1;
+ mvt *m = mvt_stack_0(config_get_history(c));
+ return (m == NULL)? -1 : mvt_get_type(m);
}
int t2(config *c) /* previous transition */
{
- if(c->mvt_array_nbelem > 1) return c->mvt_array[c->mvt_array_nbelem - 2];
- return -1;
+ mvt *m = mvt_stack_1(config_get_history(c));
+ return (m == NULL)? -1 : mvt_get_type(m);
}
int t3(config *c) /* previous transition */
{
- if(c->mvt_array_nbelem > 2) return c->mvt_array[c->mvt_array_nbelem - 3];
- return -1;
+ mvt *m = mvt_stack_2(config_get_history(c));
+ return (m == NULL)? -1 : mvt_get_type(m);
}
int t4(config *c) /* previous transition */
{
- if(c->mvt_array_nbelem > 3) return c->mvt_array[c->mvt_array_nbelem - 4];
- return -1;
+ mvt *m = mvt_stack_3(config_get_history(c));
+ return (m == NULL)? -1 : mvt_get_type(m);
}
diff --git a/maca_trans_parser/src/feat_fct.h b/maca_trans_parser/src/feat_fct.h
index e654b10cda73ab593d7fd65e2b32b11d5e012e14..457232b42e4d4f08cba39710d151faff03396e1e 100644
--- a/maca_trans_parser/src/feat_fct.h
+++ b/maca_trans_parser/src/feat_fct.h
@@ -435,7 +435,7 @@ int dist_s0_b0(config *c);
int sh(config *c);
int bh(config *c);
-int dh(config *c);
+/* int dh(config *c); */
int t1(config *c);
int t2(config *c);
diff --git a/maca_trans_parser/src/feat_lib.c b/maca_trans_parser/src/feat_lib.c
index 6342d4e602ba3cc54ce39434f156082fcf2fea31..e76cc2ab249b0db71ad1d34553ab084f9400b857 100644
--- a/maca_trans_parser/src/feat_lib.c
+++ b/maca_trans_parser/src/feat_lib.c
@@ -451,7 +451,7 @@ feat_lib *feat_lib_build(void)
feat_lib_add(fl, FEAT_TYPE_INT_8, (char *)"sh", sh);
feat_lib_add(fl, FEAT_TYPE_INT_8, (char *)"bh", bh);
- feat_lib_add(fl, FEAT_TYPE_INT_8, (char *)"dh", dh);
+ /* feat_lib_add(fl, FEAT_TYPE_INT_8, (char *)"dh", dh); */
feat_lib_add(fl, FEAT_TYPE_TRANS, (char *)"t1", t1);
feat_lib_add(fl, FEAT_TYPE_TRANS, (char *)"t2", t2);
diff --git a/maca_trans_parser/src/maca_trans_parser.c b/maca_trans_parser/src/maca_trans_parser.c
index 0b9e785b5ffb976245abaefcd4f2b10c18cb689f..6a3e6d798c83b1d8256d2adf5a3ed21fe1ee6c8b 100644
--- a/maca_trans_parser/src/maca_trans_parser.c
+++ b/maca_trans_parser/src/maca_trans_parser.c
@@ -102,7 +102,6 @@ int main(int argc, char *argv[])
ctx->features_model = feat_model_read(ctx->features_model_filename, ctx->verbose);
ctx->vocabs = dico_vec_read(ctx->vocabs_filename, ctx->hash_ratio);
- /* dico_vec_print(NULL, ctx->vocabs); */
mcd_link_to_dico(ctx->mcd_struct, ctx->vocabs, ctx->verbose);
ctx->dico_labels = dico_vec_get_dico(ctx->vocabs, (char *)"LABEL");
@@ -115,15 +114,10 @@ int main(int argc, char *argv[])
ctx->mvt_nb = ctx->dico_labels->nbelem * 2 + 3;
/* load models */
-
+
ctx->d_perceptron_features = dico_vec_get_dico(ctx->vocabs, (char *)"d_perceptron_features");
-
- if(ctx->beam_width == 1){
- simple_decoder_parser_arc_eager(ctx);
- }
- /* else
- beam_decoder(conll_file, ctx->mcd_struct, ctx->d_perceptron_features, ctx->dico_labels, ft, ctx->features_model, ctx->verbose, root_label, ctx->beam_width, ctx->mvt_nb);
- }*/
+
+ simple_decoder_parser_arc_eager(ctx);
context_free(ctx);
return 0;
diff --git a/maca_trans_parser/src/maca_trans_parser_arc_eager_mcf2cff.c b/maca_trans_parser/src/maca_trans_parser_arc_eager_mcf2cff.c
index cee41b35a5221ffb747b439f563c731495442e4f..b990c79841230ff428c9c98484b2ca777d00bd92 100644
--- a/maca_trans_parser/src/maca_trans_parser_arc_eager_mcf2cff.c
+++ b/maca_trans_parser/src/maca_trans_parser_arc_eager_mcf2cff.c
@@ -63,7 +63,7 @@ void generate_training_file_stream(FILE *output_file, context *ctx)
mcd_remove_wf_column(mcd_struct_hyp, MCD_WF_LABEL);
mcd_remove_wf_column(mcd_struct_hyp, MCD_WF_SENT_SEG);
- c = config_initial(mcf_file, mcd_struct_hyp, 5);
+ c = config_new(mcf_file, mcd_struct_hyp, 5);
while(!word_buffer_end(ref) && (sentence_nb < ctx->sent_nb)){
/*printf("************ REF ************\n");
@@ -71,12 +71,12 @@ void generate_training_file_stream(FILE *output_file, context *ctx)
printf("*****************************\n");*/
mvt_code = oracle_parser_arc_eager(c, ref, root_label);
- mvt_type = movement_type(mvt_code);
- mvt_label = movement_label(mvt_code);
+ mvt_type = movement_parser_type(mvt_code);
+ mvt_label = movement_parser_label(mvt_code);
if(ctx->debug_mode){
config_print(stdout,c);
- movement_print(stdout, mvt_code, ctx->dico_labels);
+ movement_parser_print(stdout, mvt_code, ctx->dico_labels);
}
if(ctx->trace_mode){
@@ -84,7 +84,7 @@ void generate_training_file_stream(FILE *output_file, context *ctx)
stack_print(output_file, c->st);
fprintf(output_file, "\t");
- movement_print(output_file, mvt_code, ctx->dico_labels);
+ movement_parser_print(output_file, mvt_code, ctx->dico_labels);
fprintf(output_file, "\t1\n");
}
else{
@@ -93,37 +93,37 @@ void generate_training_file_stream(FILE *output_file, context *ctx)
feat_vec_print(output_file, fv);
}
- if(mvt_type == MVT_EOS){
- movement_eos(c, 0);
+ if(mvt_type == MVT_PARSER_EOS){
+ movement_parser_eos(c);
sentence_nb++;
fprintf(stderr, "sentence %d\n", sentence_nb);
if(word_buffer_is_last(ref))
break;
}
- if(mvt_type == MVT_LEFT){
- movement_left_arc(c, mvt_label, 0);
+ if(mvt_type == MVT_PARSER_LEFT){
+ movement_parser_left_arc(c, mvt_label);
continue;
}
- if(mvt_type == MVT_RIGHT){
- movement_right_arc(c, mvt_label, 0);
+ if(mvt_type == MVT_PARSER_RIGHT){
+ movement_parser_right_arc(c, mvt_label);
word_buffer_move_right(ref);
continue;
}
- if(mvt_type == MVT_REDUCE){
- movement_reduce(c, 0);
+ if(mvt_type == MVT_PARSER_REDUCE){
+ movement_parser_reduce(c);
continue;
}
- if(mvt_type == MVT_ROOT){
- movement_root(c, 0, root_label);
+ if(mvt_type == MVT_PARSER_ROOT){
+ movement_parser_root(c, root_label);
continue;
}
- if(mvt_type == MVT_SHIFT){
- movement_shift(c, 1, 0);
+ if(mvt_type == MVT_PARSER_SHIFT){
+ movement_parser_shift(c);
word_buffer_move_right(ref);
continue;
}
diff --git a/maca_trans_parser/src/maca_trans_tagger_mcf2cff.c b/maca_trans_parser/src/maca_trans_tagger_mcf2cff.c
index 6936c6399c713da664876b45ec181e39ee043bb2..93f990a73a245950ff35a95f71b310ff5a6c2d68 100644
--- a/maca_trans_parser/src/maca_trans_tagger_mcf2cff.c
+++ b/maca_trans_parser/src/maca_trans_tagger_mcf2cff.c
@@ -12,20 +12,6 @@
#include"word_emb.h"
#include"config2feat_vec.h"
-void add_signature_to_words_in_queue(queue *bf, form2pos *f2p)
-{
- int i;
- word *w;
-
- for(i=0; i < queue_nbelem(bf); i++){
- w = queue_elt_n(bf, i);
- if(!w->signature){
- /* printf("add signature %d to word %s\n", form2pos_get_signature(f2p, w->form), w->form); */
- w->signature = form2pos_get_signature(f2p, w->form);
- }
- }
-}
-
void add_signature_to_words_in_word_buffer(word_buffer *bf, form2pos *f2p)
{
int i;
@@ -77,7 +63,7 @@ void generate_training_file_stream(FILE *output_file, context *ctx)
FILE *conll_file = myfopen(ctx->input_filename, "r");
int postag;
- c = config_initial_no_dummy_word(conll_file, ctx->mcd_struct, 5);
+ c = config_new(conll_file, ctx->mcd_struct, 5);
while(!config_is_terminal(c)){
/* config_print(stdout,c); */
@@ -101,7 +87,7 @@ void generate_training_file_buffer(FILE *output_file, context *ctx)
FILE *conll_file = myfopen(ctx->input_filename, "r");
FILE *conll_file_ref = myfopen(ctx->input_filename, "r");
int postag;
- c = config_initial(conll_file, ctx->mcd_struct, 0);
+ c = config_new(conll_file, ctx->mcd_struct, 0);
while((ref = sentence_read(conll_file_ref, ctx->mcd_struct)) && (sentence_nb < ctx->sent_nb)){
/* sentence_print(stdout, ref, NULL); */
@@ -123,7 +109,7 @@ void generate_training_file_buffer(FILE *output_file, context *ctx)
movement_tagger(c, postag, 0, 0);
}
config_free(c);
- c = config_initial(conll_file, ctx->mcd_struct, 0);
+ c = config_new(conll_file, ctx->mcd_struct, 0);
sentence_nb++;
}
}
diff --git a/maca_trans_parser/src/maca_trans_tagparser_arc_eager_mcf2cff.c b/maca_trans_parser/src/maca_trans_tagparser_arc_eager_mcf2cff.c
index e4c69d12d39df524473ca3df0b22e138c364edec..fdadaeead7901243dad7907e0d163df68f9a9705 100644
--- a/maca_trans_parser/src/maca_trans_tagparser_arc_eager_mcf2cff.c
+++ b/maca_trans_parser/src/maca_trans_tagparser_arc_eager_mcf2cff.c
@@ -12,6 +12,18 @@
#include"word_emb.h"
#include"config2feat_vec.h"
+void add_signature_to_words_in_word_buffer(word_buffer *bf, form2pos *f2p)
+{
+ int i;
+ word *w;
+
+ for(i = word_buffer_get_nbelem(bf) - 1; i >=0 ; i--){
+ w = word_buffer_get_word_n(bf, i);
+ if(word_get_signature(w) != -1) break;
+ w->signature = form2pos_get_signature(f2p, w->form);
+ }
+}
+
void maca_trans_parser_mcf2cff_help_message(context *ctx)
{
context_general_help_message(ctx);
@@ -64,7 +76,7 @@ void generate_training_file_stream(FILE *output_file, context *ctx)
mcd_remove_wf_column(mcd_struct_hyp, MCD_WF_LABEL);
mcd_remove_wf_column(mcd_struct_hyp, MCD_WF_SENT_SEG);
- c = config_initial(mcf_file, mcd_struct_hyp, 5);
+ c = config_new(mcf_file, mcd_struct_hyp, 5);
while(!word_buffer_end(ref) && (sentence_nb < ctx->sent_nb)){
/*printf("************ REF ************\n");
@@ -78,12 +90,12 @@ void generate_training_file_stream(FILE *output_file, context *ctx)
config2feat_vec_cff(ctx->features_model, c, ctx->d_perceptron_features, fv, ctx->mode);
mvt_code = oracle_tagparser_arc_eager(c, ref, root_label);
- mvt_type = movement_tagparse_type(mvt_code);
- mvt_label = movement_tagparse_label(mvt_code);
+ mvt_type = movement_tagparser_type(mvt_code);
+ mvt_label = movement_tagparser_label(mvt_code);
if(ctx->debug_mode){
config_print(stdout,c);
- movement_tagparse_print(stdout, mvt_code, ctx->dico_labels, dico_postag);
+ movement_tagparser_print(stdout, mvt_code, ctx->dico_labels, dico_postag);
fprintf(stdout, "\n");
}
@@ -92,7 +104,7 @@ void generate_training_file_stream(FILE *output_file, context *ctx)
stack_print(output_file, c->st);
fprintf(output_file, "\t");
- movement_tagparse_print(output_file, mvt_code, ctx->dico_labels, dico_postag);
+ movement_tagparser_print(output_file, mvt_code, ctx->dico_labels, dico_postag);
fprintf(output_file, "\t1\n");
}
else{
@@ -100,41 +112,41 @@ void generate_training_file_stream(FILE *output_file, context *ctx)
feat_vec_print(output_file, fv);
}
- if(mvt_type == MVT_EOS){
- movement_tagparse_eos(c, 0);
+ if(mvt_type == MVT_TAGPARSER_EOS){
+ movement_tagparser_eos(c);
sentence_nb++;
if(word_buffer_is_last(ref))
break;
}
- if(mvt_type == MVT_POSTAG){
- movement_tagparse_add_pos(c, 0, mvt_label);
+ if(mvt_type == MVT_TAGPARSER_POSTAG){
+ movement_tagparser_add_pos(c, mvt_label);
continue;
}
- if(mvt_type == MVT_LEFT){
- movement_tagparse_left_arc(c, mvt_label, 0);
+ if(mvt_type == MVT_TAGPARSER_LEFT){
+ movement_tagparser_left_arc(c, mvt_label);
continue;
}
- if(mvt_type == MVT_RIGHT){
- movement_tagparse_right_arc(c, mvt_label, 0);
+ if(mvt_type == MVT_TAGPARSER_RIGHT){
+ movement_tagparser_right_arc(c, mvt_label);
word_buffer_move_right(ref);
continue;
}
- if(mvt_type == MVT_REDUCE){
- movement_tagparse_reduce(c, 0);
+ if(mvt_type == MVT_TAGPARSER_REDUCE){
+ movement_tagparser_reduce(c);
continue;
}
- if(mvt_type == MVT_ROOT){
- movement_tagparse_root(c, 0, root_label);
+ if(mvt_type == MVT_TAGPARSER_ROOT){
+ movement_tagparser_root(c, root_label);
continue;
}
- if(mvt_type == MVT_SHIFT){
- movement_tagparse_shift(c, 1, 0);
+ if(mvt_type == MVT_TAGPARSER_SHIFT){
+ movement_tagparser_shift(c);
word_buffer_move_right(ref);
continue;
}
diff --git a/maca_trans_parser/src/movement_parser_arc_eager.c b/maca_trans_parser/src/movement_parser_arc_eager.c
index 757760eabe412fc8bbdca29f977edf5cfae68ba4..baa4c62e8f1684b296df6dde3ca69908c54f833e 100644
--- a/maca_trans_parser/src/movement_parser_arc_eager.c
+++ b/maca_trans_parser/src/movement_parser_arc_eager.c
@@ -3,127 +3,125 @@
#include<string.h>
#include"util.h"
#include"movement_parser_arc_eager.h"
+#include"movements.h"
-void movement_print(FILE *f, int mvt_code, dico *dico_labels){
- int mvt_type = movement_type(mvt_code);
- int mvt_label = movement_label(mvt_code);
+void movement_parser_print(FILE *f, int mvt_code, dico *dico_labels){
+ int mvt_type = movement_parser_type(mvt_code);
+ int mvt_label = movement_parser_label(mvt_code);
char *label;
- if(mvt_type == MVT_SHIFT) {fprintf(f, "SHIFT"); return;}
- if(mvt_type == MVT_REDUCE) {fprintf(f, "REDUCE"); return;}
- if(mvt_type == MVT_ROOT) {fprintf(f, "ROOT"); return;}
- if(mvt_type == MVT_EOS) {fprintf(f, "EOS"); return;}
- if(mvt_type == MVT_RIGHT) fprintf(f, "RIGHT");
+ if(mvt_type == MVT_PARSER_SHIFT) {fprintf(f, "SHIFT"); return;}
+ if(mvt_type == MVT_PARSER_REDUCE) {fprintf(f, "REDUCE"); return;}
+ if(mvt_type == MVT_PARSER_ROOT) {fprintf(f, "ROOT"); return;}
+ if(mvt_type == MVT_PARSER_EOS) {fprintf(f, "EOS"); return;}
+ if(mvt_type == MVT_PARSER_RIGHT) fprintf(f, "RIGHT");
else fprintf(f, "LEFT");
label = dico_int2string(dico_labels, mvt_label);
fprintf(f, " %s", label);
}
-int movement_type(int mvt)
+int movement_parser_type(int mvt)
{
- if(mvt == MVT_SHIFT) return MVT_SHIFT;
- if(mvt == MVT_REDUCE) return MVT_REDUCE;
- if(mvt == MVT_ROOT) return MVT_ROOT;
- if(mvt == MVT_EOS) return MVT_EOS;
- if(mvt % 2 == 0) return MVT_LEFT; /* even movements are left movements */
- return MVT_RIGHT; /* odd movements are right movements */
+ if(mvt == MVT_PARSER_SHIFT) return MVT_PARSER_SHIFT;
+ if(mvt == MVT_PARSER_REDUCE) return MVT_PARSER_REDUCE;
+ if(mvt == MVT_PARSER_ROOT) return MVT_PARSER_ROOT;
+ if(mvt == MVT_PARSER_EOS) return MVT_PARSER_EOS;
+ if(mvt % 2 == 0) return MVT_PARSER_LEFT; /* even movements are left movements */
+ return MVT_PARSER_RIGHT; /* odd movements are right movements */
}
-int movement_label(int mvt)
+int movement_parser_label(int mvt)
{
- if(mvt == MVT_SHIFT) return -1;
- if(mvt == MVT_REDUCE) return -1;
- if(mvt == MVT_ROOT) return -1;
- if(mvt == MVT_EOS) return -1;
+ if(mvt == MVT_PARSER_SHIFT) return -1;
+ if(mvt == MVT_PARSER_REDUCE) return -1;
+ if(mvt == MVT_PARSER_ROOT) return -1;
+ if(mvt == MVT_PARSER_EOS) return -1;
if(mvt % 2 == 0) /* even codes correspond to left movements */
return mvt / 2 - 2;
return (mvt - 1) / 2 - 2; /* odd codes correspond to right movements */
}
-int movement_eos(config *c, float score)
+int movement_parser_eos(config *c)
{
- if(stack_is_empty(config_get_stack(c))) return 0;
-
- /* word on the top of the stack is sent_seg */
- word_set_sent_seg(stack_top(config_get_stack(c)), 1);
-
- /* (config_get_stack(c))->top = 0; */
-
- config_add_mvt(c, MVT_EOS);
- return 1;
+ return movement_eos(c, MVT_PARSER_EOS);
}
-int movement_left_arc(config *c, int label, float score)
+int movement_parser_eos_undo(config *c)
{
- if(stack_is_empty(config_get_stack(c))) return 0;
- /* if(word_buffer_is_empty(config_get_buffer(c))) return 0; */
-
- /* word on top of the stack should not have a governor */
- if(word_get_gov(stack_top(config_get_stack(c))) != WORD_INVALID_GOV) return 0;
+ return movement_eos_undo(c);
+}
- word *gov = word_buffer_b0(config_get_buffer(c));
- word *dep = stack_top(config_get_stack(c));
- int dist = (word_get_index(gov)) - (word_get_index(dep));
+int movement_parser_left_arc(config *c, int label)
+{
+ return movement_left_arc(c, movement_parser_left_code(label), label);
+}
- /* create a new dependency */
- word_set_gov(dep, dist);
- word_set_label(dep, label);
+int movement_parser_left_arc_undo(config *c)
+{
+ return movement_left_arc_undo(c);
+}
- stack_pop(config_get_stack(c));
- config_add_mvt(c, movement_left_code(label));
- return 1;
+int movement_parser_right_arc(config *c, int label)
+{
+ return movement_right_arc(c, movement_parser_right_code(label), label);
}
-int movement_right_arc(config *c, int label, float score)
+int movement_parser_right_arc_undo(config *c)
{
- if(stack_is_empty(config_get_stack(c))) return 0;
-
- word *gov = stack_top(config_get_stack(c));
- word *dep = word_buffer_b0(config_get_buffer(c));
- int dist = (word_get_index(gov)) - (word_get_index(dep));
-
- /* create a new dependency */
- word_set_gov(dep, dist);
- word_set_label(dep, label);
-
- stack_push(config_get_stack(c), word_buffer_b0(config_get_buffer(c)));
- word_buffer_move_right(config_get_buffer(c));
-
- config_add_mvt(c, movement_right_code(label));
- return 1;
-}
-
-int movement_shift(config *c, int stream, float score)
+ return movement_right_arc_undo(c);
+}
+
+int movement_parser_shift(config *c)
{
- if(word_buffer_is_empty(config_get_buffer(c))) return 0;
- stack_push(config_get_stack(c), word_buffer_b0(config_get_buffer(c)));
- word_buffer_move_right(config_get_buffer(c));
- config_add_mvt(c, MVT_SHIFT);
- return 1;
+ return movement_shift(c, MVT_PARSER_SHIFT);
}
-int movement_reduce(config *c, float score)
+int movement_parser_shift_undo(config *c)
{
- if(stack_nbelem(config_get_stack(c)) <= 1) return 0;
+ return movement_shift_undo(c);
+}
- /* if(stack_is_empty(config_get_stack(c))) return 0; */
+int movement_parser_reduce(config *c)
+{
+ return movement_reduce(c, MVT_PARSER_REDUCE);
+}
- /* word on top of stack must have a governor */
- if(word_get_gov(stack_top(config_get_stack(c))) == WORD_INVALID_GOV) return 0;
- stack_pop(config_get_stack(c));
- config_add_mvt(c, MVT_REDUCE);
- return 1;
+int movement_parser_reduce_undo(config *c)
+{
+ return movement_reduce_undo(c);
}
-int movement_root(config *c, float score, int root_code)
+int movement_parser_root(config *c, int root_code)
{
- word *s0 = stack_top(config_get_stack(c));
- if(s0 == NULL) return 0;
- word_set_gov(s0, 0);
- word_set_label(s0, root_code);
- s0->is_root = 1;
+ return movement_root(c, MVT_PARSER_ROOT, root_code);
+}
- stack_pop(config_get_stack(c));
+int movement_parser_root_undo(config *c)
+{
+ return movement_root_undo(c);
+}
- config_add_mvt(c, MVT_ROOT);
- return 1;
+int movement_parser_undo(config *c)
+{
+ int result;
+ int mvt_type = mvt_get_type(mvt_stack_top(config_get_history(c)));
+ switch(mvt_type){
+ case MVT_PARSER_LEFT :
+ result = movement_parser_left_arc_undo(c);
+ break;
+ case MVT_PARSER_RIGHT:
+ result = movement_parser_right_arc_undo(c);
+ break;
+ case MVT_PARSER_REDUCE:
+ result = movement_parser_reduce_undo(c);
+ break;
+ case MVT_PARSER_ROOT:
+ result = movement_parser_root_undo(c);
+ break;
+ case MVT_PARSER_EOS:
+ result = movement_parser_eos_undo(c);
+ break;
+ case MVT_PARSER_SHIFT:
+ result = movement_parser_shift_undo(c);
+ }
+ return result;
}
diff --git a/maca_trans_parser/src/movement_parser_arc_eager.h b/maca_trans_parser/src/movement_parser_arc_eager.h
index 63f2f8a65f5921ecfde4594532bda3b9feb6384e..2f4205931757b78b9bb9f89df997a32e39fdebb5 100644
--- a/maca_trans_parser/src/movement_parser_arc_eager.h
+++ b/maca_trans_parser/src/movement_parser_arc_eager.h
@@ -2,32 +2,37 @@
#define __MOVEMENT_PARSER_ARC_EAGER__
#include"config.h"
-#include"feat_vec.h"
-
-
-#define MVT_SHIFT 0
-#define MVT_REDUCE 1
-#define MVT_ROOT 2
-#define MVT_EOS 3
-#define MVT_LEFT 4
-#define MVT_RIGHT 5
+#include"dico.h"
+#define MVT_PARSER_SHIFT 0
+#define MVT_PARSER_REDUCE 1
+#define MVT_PARSER_ROOT 2
+#define MVT_PARSER_EOS 3
+#define MVT_PARSER_LEFT 4
+#define MVT_PARSER_RIGHT 5
/* even movements are left movements (except 0, which is shift and 2 which is root) */
-#define movement_left_code(label) (2 * (label) + 4)
+#define movement_parser_left_code(label) (2 * (label) + 4)
/* odd movements are right movements (except 1, which is reduce and 3 which is end_of_sentence) */
-#define movement_right_code(label) (2 * (label) + 5)
-
-int movement_type(int mvt);
-int movement_label(int mvt);
-
-int movement_left_arc(config *c, int label, float score);
-int movement_right_arc(config *c, int label, float score);
-int movement_shift(config *c, int stream, float score);
-int movement_reduce(config *c, float score);
-int movement_root(config *c, float score, int root_code);
-int movement_eos(config *c, float score);
-void movement_print(FILE *f, int mvt_code, dico *dico_labels);
+#define movement_parser_right_code(label) (2 * (label) + 5)
+
+int movement_parser_type(int mvt);
+int movement_parser_label(int mvt);
+
+int movement_parser_left_arc(config *c, int label);
+int movement_parser_left_arc_undo(config *c);
+int movement_parser_right_arc(config *c, int label);
+int movement_parser_right_arc_undo(config *c);
+int movement_parser_shift(config *c);
+int movement_parser_shift_undo(config *c);
+int movement_parser_reduce(config *c);
+int movement_parser_reduce_undo(config *c);
+int movement_parser_root(config *c, int root_code);
+int movement_parser_root_undo(config *c);
+int movement_parser_eos(config *c);
+int movement_parser_eos_undo(config *c);
+int movement_parser_undo(config *c);
+void movement_parser_print(FILE *f, int mvt_code, dico *dico_labels);
#endif
diff --git a/maca_trans_parser/src/movement_tagparser_arc_eager.c b/maca_trans_parser/src/movement_tagparser_arc_eager.c
index df5f5a730cf2e22032dfff3e427bccd4a13ce6d4..b7e494bcabc72e7a51e3bc1a91619722096506e5 100644
--- a/maca_trans_parser/src/movement_tagparser_arc_eager.c
+++ b/maca_trans_parser/src/movement_tagparser_arc_eager.c
@@ -2,48 +2,49 @@
#include<stdlib.h>
#include<string.h>
#include"util.h"
+#include"movements.h"
#include"movement_tagparser_arc_eager.h"
-void movement_tagparse_print(FILE *f, int mvt_code, dico *dico_labels, dico *dico_postag)
+void movement_tagparser_print(FILE *f, int mvt_code, dico *dico_labels, dico *dico_postag)
{
- int mvt_type = movement_tagparse_type(mvt_code);
- int mvt_label = movement_tagparse_label(mvt_code);
+ int mvt_type = movement_tagparser_type(mvt_code);
+ int mvt_label = movement_tagparser_label(mvt_code);
char *label;
- if(mvt_type == MVT_SHIFT) {fprintf(f, "SHIFT"); return;}
- if(mvt_type == MVT_REDUCE) {fprintf(f, "REDUCE"); return;}
- if(mvt_type == MVT_ROOT) {fprintf(f, "ROOT"); return;}
- if(mvt_type == MVT_EOS) {fprintf(f, "EOS"); return;}
- if(mvt_type == MVT_POSTAG){
+ if(mvt_type == MVT_TAGPARSER_SHIFT) {fprintf(f, "SHIFT"); return;}
+ if(mvt_type == MVT_TAGPARSER_REDUCE) {fprintf(f, "REDUCE"); return;}
+ if(mvt_type == MVT_TAGPARSER_ROOT) {fprintf(f, "ROOT"); return;}
+ if(mvt_type == MVT_TAGPARSER_EOS) {fprintf(f, "EOS"); return;}
+ if(mvt_type == MVT_TAGPARSER_POSTAG){
fprintf(f, "POSTAG");
label = dico_int2string(dico_postag, mvt_label);
fprintf(f, " %s", label);
return;
}
- if(mvt_type == MVT_RIGHT) fprintf(f, "RIGHT");
+ if(mvt_type == MVT_TAGPARSER_RIGHT) fprintf(f, "RIGHT");
else fprintf(f, "LEFT");
label = dico_int2string(dico_labels, mvt_label);
fprintf(f, " %s", label);
}
-int movement_tagparse_type(int mvt)
+int movement_tagparser_type(int mvt)
{
- if(mvt == MVT_SHIFT) return MVT_SHIFT; /* 0 */
- if(mvt == MVT_REDUCE) return MVT_REDUCE; /* 1 */
- if(mvt == MVT_ROOT) return MVT_ROOT; /* 2 */
- if(mvt == MVT_EOS) return MVT_EOS; /* 3 */
- if(mvt % 3 == 0) return MVT_RIGHT; /* 4, 7, 10 ... */
- if(mvt % 3 == 1) return MVT_POSTAG; /* 5, 8, 11 ... */
- /*if(mvt % 3 == 2)*/ return MVT_LEFT; /* 6, 9, 12 ... */
+ if(mvt == MVT_TAGPARSER_SHIFT) return MVT_TAGPARSER_SHIFT; /* 0 */
+ if(mvt == MVT_TAGPARSER_REDUCE) return MVT_TAGPARSER_REDUCE; /* 1 */
+ if(mvt == MVT_TAGPARSER_ROOT) return MVT_TAGPARSER_ROOT; /* 2 */
+ if(mvt == MVT_TAGPARSER_EOS) return MVT_TAGPARSER_EOS; /* 3 */
+ if(mvt % 3 == 0) return MVT_TAGPARSER_RIGHT; /* 4, 7, 10 ... */
+ if(mvt % 3 == 1) return MVT_TAGPARSER_POSTAG; /* 5, 8, 11 ... */
+ /*if(mvt % 3 == 2)*/ return MVT_TAGPARSER_LEFT; /* 6, 9, 12 ... */
}
-int movement_tagparse_label(int mvt)
+int movement_tagparser_label(int mvt)
{
- if(mvt == MVT_SHIFT) return -1;
- if(mvt == MVT_REDUCE) return -1;
- if(mvt == MVT_ROOT) return -1;
- if(mvt == MVT_EOS) return -1;
+ if(mvt == MVT_TAGPARSER_SHIFT) return -1;
+ if(mvt == MVT_TAGPARSER_REDUCE) return -1;
+ if(mvt == MVT_TAGPARSER_ROOT) return -1;
+ if(mvt == MVT_TAGPARSER_EOS) return -1;
if(mvt % 3 == 1) /* pos movement */
return (mvt - 4) / 3;
if(mvt % 3 == 2) /* left movement */
@@ -52,106 +53,101 @@ int movement_tagparse_label(int mvt)
return (mvt - 6) / 3;
}
-int movement_tagparse_add_pos(config *c, float score, int pos)
+int movement_tagparser_add_pos(config *c, int pos)
{
- if(word_buffer_b0(config_get_buffer(c)) == NULL) return 0;
- if(word_get_pos(word_buffer_b0(config_get_buffer(c))) != -1) return 0;
- word_set_pos(word_buffer_b0(config_get_buffer(c)), pos);
-
- /* stack_push(config_get_stack(c), word_buffer_b0(config_get_buffer(c)));
- word_buffer_move_right(config_get_buffer(c));*/
- config_add_mvt(c, movement_tagparse_postag(pos));
-
- return 1;
+ return movement_add_pos(c, movement_tagparser_postag_code(pos), pos);
}
-int movement_tagparse_eos(config *c, float score)
+int movement_tagparser_add_pos_undo(config *c)
{
- if(stack_is_empty(config_get_stack(c))) return 0;
- if(word_get_sent_seg(stack_top(config_get_stack(c))) == 1) return 0;
-
-
- /* word on the top of the stack is sent_seg */
- word_set_sent_seg(stack_top(config_get_stack(c)), 1);
-
- /* (config_get_stack(c))->top = 0; */
-
- config_add_mvt(c, MVT_EOS);
- return 1;
+ return movement_add_pos_undo(c);
}
-int movement_tagparse_left_arc(config *c, int label, float score)
+int movement_tagparser_eos(config *c)
{
- if(stack_is_empty(config_get_stack(c))) return 0;
- /* if(word_buffer_is_empty(config_get_buffer(c))) return 0; */
-
- /* word on top of the stack should not have a governor */
- if(word_get_gov(stack_top(config_get_stack(c))) != WORD_INVALID_GOV) return 0;
+ return movement_eos(c, MVT_TAGPARSER_EOS);
+}
- word *gov = word_buffer_b0(config_get_buffer(c));
- word *dep = stack_top(config_get_stack(c));
- int dist = (word_get_index(gov)) - (word_get_index(dep));
+int movement_tagparser_eos_undo(config *c)
+{
+ return movement_eos_undo(c);
+}
- /* create a new dependency */
- word_set_gov(dep, dist);
- word_set_label(dep, label);
+int movement_tagparser_left_arc(config *c, int label)
+{
+ return movement_left_arc(c, movement_tagparser_left_code(label), label);
+}
- stack_pop(config_get_stack(c));
- config_add_mvt(c, movement_tagparse_left_code(label));
- return 1;
+int movement_tagparser_left_arc_undo(config *c)
+{
+ return movement_left_arc_undo(c);
}
-int movement_tagparse_right_arc(config *c, int label, float score)
+int movement_tagparser_right_arc(config *c, int label)
{
- if(stack_is_empty(config_get_stack(c))) return 0;
-
- word *gov = stack_top(config_get_stack(c));
- word *dep = word_buffer_b0(config_get_buffer(c));
- int dist = (word_get_index(gov)) - (word_get_index(dep));
+ return movement_right_arc(c, movement_tagparser_right_code(label), label);
+}
- /* create a new dependency */
- word_set_gov(dep, dist);
- word_set_label(dep, label);
-
- stack_push(config_get_stack(c), word_buffer_b0(config_get_buffer(c)));
- word_buffer_move_right(config_get_buffer(c));
-
- config_add_mvt(c, movement_tagparse_right_code(label));
- return 1;
+int movement_tagparser_right_arc_undo(config *c)
+{
+ return movement_right_arc_undo(c);
}
-int movement_tagparse_shift(config *c, int stream, float score)
+int movement_tagparser_shift(config *c)
{
- if(word_buffer_is_empty(config_get_buffer(c))) return 0;
- stack_push(config_get_stack(c), word_buffer_b0(config_get_buffer(c)));
- word_buffer_move_right(config_get_buffer(c));
- config_add_mvt(c, MVT_SHIFT);
- return 1;
+ return movement_shift(c, MVT_TAGPARSER_SHIFT);
}
-int movement_tagparse_reduce(config *c, float score)
+int movement_tagparser_shift_undo(config *c)
{
- if(stack_nbelem(config_get_stack(c)) <= 1) return 0;
+ return movement_shift_undo(c);
+}
- /* if(stack_is_empty(config_get_stack(c))) return 0; */
+int movement_tagparser_reduce(config *c)
+{
+ return movement_reduce(c, MVT_TAGPARSER_REDUCE);
+}
- /* word on top of stack must have a governor */
- if(word_get_gov(stack_top(config_get_stack(c))) == WORD_INVALID_GOV) return 0;
- stack_pop(config_get_stack(c));
- config_add_mvt(c, MVT_REDUCE);
- return 1;
+int movement_tagparser_reduce_undo(config *c)
+{
+ return movement_reduce_undo(c);
}
-int movement_tagparse_root(config *c, float score, int root_code)
+int movement_tagparser_root(config *c, int root_code)
{
- word *s0 = stack_top(config_get_stack(c));
- if(s0 == NULL) return 0;
- word_set_gov(s0, 0);
- word_set_label(s0, root_code);
- s0->is_root = 1;
+ return movement_root(c, MVT_TAGPARSER_ROOT, root_code);
+}
- stack_pop(config_get_stack(c));
+int movement_tagparser_root_undo(config *c)
+{
+ return movement_root_undo(c);
+}
- config_add_mvt(c, MVT_ROOT);
- return 1;
+int movement_tagparser_undo(config *c)
+{
+ int result;
+ int mvt_type = mvt_get_type(mvt_stack_top(config_get_history(c)));
+ switch(mvt_type){
+ case MVT_TAGPARSER_POSTAG :
+ result = movement_tagparser_left_arc_undo(c);
+ break;
+ case MVT_TAGPARSER_LEFT :
+ result = movement_tagparser_left_arc_undo(c);
+ break;
+ case MVT_TAGPARSER_RIGHT:
+ result = movement_tagparser_right_arc_undo(c);
+ break;
+ case MVT_TAGPARSER_REDUCE:
+ result = movement_tagparser_reduce_undo(c);
+ break;
+ case MVT_TAGPARSER_ROOT:
+ result = movement_tagparser_root_undo(c);
+ break;
+ case MVT_TAGPARSER_EOS:
+ result = movement_tagparser_eos_undo(c);
+ break;
+ case MVT_TAGPARSER_SHIFT:
+ result = movement_tagparser_shift_undo(c);
+ }
+ return result;
}
diff --git a/maca_trans_parser/src/movement_tagparser_arc_eager.h b/maca_trans_parser/src/movement_tagparser_arc_eager.h
index 89f5a681aeb1f36db1dd1924d06c3cfa3c0543ea..98e42c48652968e9375519e32e9f51f83e79165b 100644
--- a/maca_trans_parser/src/movement_tagparser_arc_eager.h
+++ b/maca_trans_parser/src/movement_tagparser_arc_eager.h
@@ -2,34 +2,33 @@
#define __MOVEMENT_TAGPARSER_ARC_EAGER__
#include"config.h"
-#include"feat_vec.h"
+#include"dico.h"
+#define MVT_TAGPARSER_SHIFT 0
+#define MVT_TAGPARSER_REDUCE 1
+#define MVT_TAGPARSER_ROOT 2
+#define MVT_TAGPARSER_EOS 3
+#define MVT_TAGPARSER_LEFT 4
+#define MVT_TAGPARSER_RIGHT 5
+#define MVT_TAGPARSER_POSTAG 6
-#define MVT_SHIFT 0
-#define MVT_REDUCE 1
-#define MVT_ROOT 2
-#define MVT_EOS 3
-#define MVT_LEFT 4
-#define MVT_RIGHT 5
-#define MVT_POSTAG 6
-
-#define movement_tagparse_postag(postag) (3 * (postag) + 4)
+#define movement_tagparser_postag_code(postag) (3 * (postag) + 4)
/* even movements are left movements (except 0, which is shift and 2 which is root) */
-#define movement_tagparse_left_code(label) (3 * (label) + 5)
+#define movement_tagparser_left_code(label) (3 * (label) + 5)
/* odd movements are right movements (except 1, which is reduce and 3 which is end_of_sentence) */
-#define movement_tagparse_right_code(label) (3 * (label) + 6)
-
-int movement_tagparse_type(int mvt);
-int movement_tagparse_label(int mvt);
-
-int movement_tagparse_left_arc(config *c, int label, float score);
-int movement_tagparse_right_arc(config *c, int label, float score);
-int movement_tagparse_shift(config *c, int stream, float score);
-int movement_tagparse_reduce(config *c, float score);
-int movement_tagparse_root(config *c, float score, int root_code);
-int movement_tagparse_eos(config *c, float score);
-int movement_tagparse_add_pos(config *c, float score, int postag);
-void movement_tagparse_print(FILE *f, int mvt_code, dico *dico_labels, dico *dico_postag);
+#define movement_tagparser_right_code(label) (3 * (label) + 6)
+
+int movement_tagparser_type(int mvt);
+int movement_tagparser_label(int mvt);
+
+int movement_tagparser_left_arc(config *c, int label);
+int movement_tagparser_right_arc(config *c, int label);
+int movement_tagparser_shift(config *c);
+int movement_tagparser_reduce(config *c);
+int movement_tagparser_root(config *c, int root_code);
+int movement_tagparser_eos(config *c);
+int movement_tagparser_add_pos(config *c, int postag);
+void movement_tagparser_print(FILE *f, int mvt_code, dico *dico_labels, dico *dico_postag);
#endif
diff --git a/maca_trans_parser/src/movements.c b/maca_trans_parser/src/movements.c
new file mode 100644
index 0000000000000000000000000000000000000000..cccdf083fc2f299f2998f8679968f2c4db3758df
--- /dev/null
+++ b/maca_trans_parser/src/movements.c
@@ -0,0 +1,183 @@
+#include<stdio.h>
+#include<stdlib.h>
+#include<string.h>
+#include"util.h"
+#include"movements.h"
+
+int movement_eos(config *c, int movement_code)
+{
+ if(stack_is_empty(config_get_stack(c))) return 0;
+ word *s0 = stack_top(config_get_stack(c));
+
+ /* word on the top of the stack is sent_seg */
+ word_set_sent_seg(s0, 1);
+
+ config_push_mvt(c, movement_code, s0, NULL);
+ return 1;
+}
+
+int movement_eos_undo(config *c)
+{
+ word *gov = mvt_get_gov(mvt_stack_top(config_get_history(c)));
+
+ /* word on the top of the stack is not sent_seg anymore */
+ word_set_sent_seg(gov, 0);
+
+ mvt_free(config_pop_mvt(c));
+ return 1;
+}
+
+int movement_left_arc(config *c, int movement_code, int label)
+{
+ if(stack_is_empty(config_get_stack(c))) return 0;
+ /* if(word_buffer_is_empty(config_get_buffer(c))) return 0; */
+
+ /* word on top of the stack should not have a governor */
+ if(word_get_gov(stack_top(config_get_stack(c))) != WORD_INVALID_GOV) return 0;
+
+ word *gov = word_buffer_b0(config_get_buffer(c));
+ word *dep = stack_top(config_get_stack(c));
+ int dist = (word_get_index(gov)) - (word_get_index(dep));
+
+ /* create a new dependency */
+ word_set_gov(dep, dist);
+ word_set_label(dep, label);
+
+ config_push_mvt(c, movement_code, gov, dep);
+
+ stack_pop(config_get_stack(c));
+ return 1;
+}
+
+int movement_left_arc_undo(config *c)
+{
+
+ word *dep = mvt_get_dep(mvt_stack_top(config_get_history(c)));
+
+ /* undo dependency */
+ word_set_gov(dep, WORD_INVALID_GOV);
+ word_set_label(dep, -1);
+
+ mvt_free(config_pop_mvt(c));
+
+ stack_push(config_get_stack(c), dep);
+ return 1;
+}
+
+
+int movement_right_arc(config *c, int movement_code, int label)
+{
+ if(stack_is_empty(config_get_stack(c))) return 0;
+
+ word *gov = stack_top(config_get_stack(c));
+ word *dep = word_buffer_b0(config_get_buffer(c));
+ int dist = (word_get_index(gov)) - (word_get_index(dep));
+
+ /* create a new dependency */
+ word_set_gov(dep, dist);
+ word_set_label(dep, label);
+
+ stack_push(config_get_stack(c), word_buffer_b0(config_get_buffer(c)));
+ word_buffer_move_right(config_get_buffer(c));
+
+ config_push_mvt(c, movement_code, gov, dep);
+ return 1;
+}
+
+int movement_right_arc_undo(config *c)
+{
+ word *dep = mvt_get_dep(mvt_stack_top(config_get_history(c)));
+
+ /* undo dependency */
+ word_set_gov(dep, WORD_INVALID_GOV);
+ word_set_label(dep, -1);
+
+ stack_pop(config_get_stack(c));
+ word_buffer_move_left(config_get_buffer(c));
+ mvt_free(config_pop_mvt(c));
+ return 1;
+}
+
+int movement_shift(config *c, int movement_code)
+{
+ if(word_buffer_is_empty(config_get_buffer(c))) return 0;
+ word *b0 = word_buffer_b0(config_get_buffer(c));
+ stack_push(config_get_stack(c), b0);
+ config_push_mvt(c, movement_code, b0, NULL);
+ word_buffer_move_right(config_get_buffer(c));
+ return 1;
+}
+
+int movement_shift_undo(config *c)
+{
+ stack_pop(config_get_stack(c));
+ mvt_free(config_pop_mvt(c));
+ word_buffer_move_left(config_get_buffer(c));
+ return 1;
+}
+
+int movement_reduce(config *c, int movement_code)
+{
+ if(stack_nbelem(config_get_stack(c)) <= 1) return 0;
+ word *s0 = stack_top(config_get_stack(c));
+ /* word on top of stack must have a governor */
+ if(word_get_gov(s0) == WORD_INVALID_GOV) return 0;
+
+ config_push_mvt(c, movement_code, s0, NULL);
+ stack_pop(config_get_stack(c));
+ return 1;
+}
+
+int movement_reduce_undo(config *c)
+{
+ word *gov = mvt_get_gov(mvt_stack_top(config_get_history(c)));
+ mvt_free(config_pop_mvt(c));
+ stack_push(config_get_stack(c), gov);
+ return 1;
+}
+
+int movement_root(config *c, int movement_code, int root_code)
+{
+ word *s0 = stack_top(config_get_stack(c));
+ if(s0 == NULL) return 0;
+ word_set_gov(s0, 0);
+ word_set_label(s0, root_code);
+ s0->is_root = 1;
+
+ stack_pop(config_get_stack(c));
+
+ config_push_mvt(c, movement_code, s0, NULL);
+ return 1;
+}
+
+int movement_root_undo(config *c)
+{
+ word *gov = mvt_get_gov(mvt_stack_top(config_get_history(c)));
+ if(gov == NULL) return 0;
+ word_set_gov(gov, WORD_INVALID_GOV);
+ word_set_label(gov, -1);
+ gov->is_root = 0;
+
+ stack_push(config_get_stack(c), gov);
+ mvt_free(config_pop_mvt(c));
+ return 1;
+}
+
+int movement_add_pos(config *c, int movement_code, int pos)
+{
+ if(word_buffer_b0(config_get_buffer(c)) == NULL) return 0;
+ word *b0 = word_buffer_b0(config_get_buffer(c));
+ if(word_get_pos(b0) != -1) return 0;
+ word_set_pos(b0, pos);
+ config_push_mvt(c, movement_code, b0, NULL);
+ return 1;
+}
+
+int movement_add_pos_undo(config *c)
+{
+ word *gov = mvt_get_gov(mvt_stack_top(config_get_history(c)));
+ word_set_pos(gov, -1);
+ mvt_free(config_pop_mvt(c));
+ return 1;
+}
+
diff --git a/maca_trans_parser/src/movements.h b/maca_trans_parser/src/movements.h
new file mode 100644
index 0000000000000000000000000000000000000000..9d8adf55ff082cecf9cf32609b12e84372c19991
--- /dev/null
+++ b/maca_trans_parser/src/movements.h
@@ -0,0 +1,25 @@
+#ifndef __MOVEMENTS_
+#define __MOVEMENTS_
+
+#include<stdio.h>
+#include<stdlib.h>
+#include<string.h>
+#include"config.h"
+
+int movement_eos(config *c, int movement_code);
+int movement_eos_undo(config *c);
+int movement_left_arc(config *c, int movement_code, int label);
+int movement_left_arc_undo(config *c);
+int movement_right_arc(config *c, int movement_code, int label);
+int movement_right_arc_undo(config *c);
+int movement_shift(config *c, int movement_code);
+int movement_shift_undo(config *c);
+int movement_reduce(config *c, int movement_code);
+int movement_reduce_undo(config *c);
+int movement_root(config *c, int movement_code, int root_code);
+int movement_root_undo(config *c);
+int movement_undo(config *c);
+int movement_add_pos(config *c, int movement_code, int pos);
+int movement_add_pos_undo(config *c);
+
+#endif
diff --git a/maca_trans_parser/src/mvt.c b/maca_trans_parser/src/mvt.c
index 228bfbf546b39a6aa1d30ecf22813bc9f565c446..fd5c549fc2561777112c2d0714a9406d5847b119 100644
--- a/maca_trans_parser/src/mvt.c
+++ b/maca_trans_parser/src/mvt.c
@@ -3,7 +3,7 @@
#include"util.h"
#include"mvt.h"
-mvt *mvt_new(int type, int gov, int dep)
+mvt *mvt_new(int type, word *gov, word *dep)
{
mvt *m = memalloc(sizeof(mvt));
m->type = type;
@@ -21,6 +21,6 @@ void mvt_free(mvt *m)
void mvt_print(FILE *f, mvt *m)
{
if(m)
- printf("type = %d gov = %d dep = %d\n", mvt_get_type(m), mvt_get_gov(m), mvt_get_dep(m));
+ printf("type = %d gov = %d dep = %d\n", mvt_get_type(m), word_get_index(mvt_get_gov(m)), word_get_index(mvt_get_dep(m)));
}
diff --git a/maca_trans_parser/src/mvt.h b/maca_trans_parser/src/mvt.h
index 863fb0d9a9080a0139f22789b6fd3e1640d71a76..97ca8f50526ab072f94f77678e116fba003ccedf 100644
--- a/maca_trans_parser/src/mvt.h
+++ b/maca_trans_parser/src/mvt.h
@@ -3,6 +3,7 @@
#include<stdio.h>
#include<stdlib.h>
+#include"word.h"
#define mvt_get_type(m) ((m)->type)
#define mvt_set_type(m,v) ((m)->type = (v))
@@ -15,11 +16,11 @@
typedef struct {
int type; /* type of the movement */
- int gov; /* first argument of the movement (governor if it is a dependency creation movement) */
- int dep; /* second argument of the movement (dependent if it is a dependency creation movement) */
+ word* gov; /* first argument of the movement (governor if it is a dependency creation movement) */
+ word* dep; /* second argument of the movement (dependent if it is a dependency creation movement) */
} mvt;
-mvt *mvt_new(int type, int gov, int dep);
+mvt *mvt_new(int type, word* gov, word* dep);
void mvt_free(mvt *m);
void mvt_print(FILE *f, mvt *m);
diff --git a/maca_trans_parser/src/mvt_stack.c b/maca_trans_parser/src/mvt_stack.c
index 2e301ecb8a3f0144e6def1c8599d0ac6cda3986a..6f99a119ac3f662062cadf6545cbad3fdd0c33bd 100644
--- a/maca_trans_parser/src/mvt_stack.c
+++ b/maca_trans_parser/src/mvt_stack.c
@@ -6,8 +6,8 @@
void mvt_stack_print(FILE *f, mvt_stack *ms)
{
int i;
-for(i=0; i < mvt_stack_nbelem(ms); i++)
- mvt_print(f, mvt_stack_elt_n(ms, i));
+ for(i=0; i < mvt_stack_nbelem(ms); i++)
+ mvt_print(f, mvt_stack_elt_n(ms, i));
}
void mvt_stack_free(mvt_stack *s)
@@ -35,7 +35,4 @@ void mvt_stack_push(mvt_stack *s, mvt *m)
s->top++;
}
-mvt_stack *mvt_stack_copy(mvt_stack *ms)
-{}
-
diff --git a/maca_trans_parser/src/mvt_stack.h b/maca_trans_parser/src/mvt_stack.h
index 5648fc7ee0e5a9d91362bd522976504bce678110..5c66e8c2e2aa78406d644544587436672ed791d8 100644
--- a/maca_trans_parser/src/mvt_stack.h
+++ b/maca_trans_parser/src/mvt_stack.h
@@ -3,30 +3,26 @@
#include"mvt.h"
-#define mvt_stack_nbelem(s) ((s)->top)
-#define mvt_stack_elt_n(s, n) ((s)->array[(s)->top - (n) - 1])
-#define mvt_stack_pop(s) (((s)->top == 0)? NULL : (s)->array[--((s)->top)])
-
-
-#define mvt_stack_is_empty(s) (((s)->top == 0) ? 1 : 0)
-
-
-#define mvt_stack_top(s) (((s)->top > 0)? (s)->array[(s)->top - 1] : NULL)
-#define mvt_stack_0(s) (((s)->top > 0)? (s)->array[(s)->top - 1] : NULL)
-#define mvt_stack_1(s) (((s)->top > 1)? (s)->array[(s)->top - 2] : NULL)
-#define mvt_stack_2(s) (((s)->top > 2)? (s)->array[(s)->top - 3] : NULL)
-#define mvt_stack_3(s) (((s)->top > 3)? (s)->array[(s)->top - 4] : NULL)
-#define mvt_stack_4(s) (((s)->top > 4)? (s)->array[(s)->top - 5] : NULL)
-#define mvt_stack_5(s) (((s)->top > 5)? (s)->array[(s)->top - 6] : NULL)
-
typedef struct {
int size;
mvt **array;
int top;
} mvt_stack;
+
+#define mvt_stack_nbelem(s) ((s)->top)
+#define mvt_stack_elt_n(s, n) ((s)->array[(s)->top - (n) - 1])
+#define mvt_stack_pop(s) (((s)->top == 0)? NULL : (s)->array[--((s)->top)])
+#define mvt_stack_is_empty(s) (((s)->top == 0) ? 1 : 0)
+#define mvt_stack_top(s) (((s)->top > 0)? (s)->array[(s)->top - 1] : NULL)
+#define mvt_stack_0(s) (((s)->top > 0)? (s)->array[(s)->top - 1] : NULL)
+#define mvt_stack_1(s) (((s)->top > 1)? (s)->array[(s)->top - 2] : NULL)
+#define mvt_stack_2(s) (((s)->top > 2)? (s)->array[(s)->top - 3] : NULL)
+#define mvt_stack_3(s) (((s)->top > 3)? (s)->array[(s)->top - 4] : NULL)
+#define mvt_stack_4(s) (((s)->top > 4)? (s)->array[(s)->top - 5] : NULL)
+#define mvt_stack_5(s) (((s)->top > 5)? (s)->array[(s)->top - 6] : NULL)
+
mvt_stack *mvt_stack_new(void);
-mvt_stack *mvt_stack_copy(mvt_stack *ms);
void mvt_stack_push(mvt_stack *ms, mvt *m);
void mvt_stack_print(FILE *f, mvt_stack *ms);
void mvt_stack_free(mvt_stack *ms);
diff --git a/maca_trans_parser/src/oracle_parser_arc_eager.c b/maca_trans_parser/src/oracle_parser_arc_eager.c
index 93d662b23fb97f5f03ea3cbd5527c08746e4a574..074400e2b14b8bede8cbef8815224b2a73553b95 100644
--- a/maca_trans_parser/src/oracle_parser_arc_eager.c
+++ b/maca_trans_parser/src/oracle_parser_arc_eager.c
@@ -11,27 +11,6 @@ int check_all_dependents_of_word_in_ref_are_in_hyp(config *c, word_buffer *ref,
int gov_hyp;
int sentence_change;
-#if 0
- for(dep = word_index - 1; (dep >= 0) && (word_get_sent_seg(word_buffer_get_word_n(ref, dep)) == 0); dep--){
- gov_ref = word_get_gov_index(word_buffer_get_word_n(ref, dep));
- if(gov_ref == word_index){ /* dep is a dependent of word in ref */
- /* check that dep has the same governor in hyp */
- gov_hyp = word_get_gov_index(word_buffer_get_word_n(config_get_buffer(c), dep));
- if(gov_hyp != gov_ref) return 0;
- }
- }
-
- for(dep = word_index + 1; ((dep < word_buffer_get_nbelem(ref)) && (word_get_sent_seg(word_buffer_get_word_n(ref, dep)) == 0)); dep++){
- gov_ref = word_get_gov_index(word_buffer_get_word_n(ref, dep));
- if(gov_ref == word_index){ /* dep is a dependent of word in ref */
- /* check that dep has the same governor in hyp */
- gov_hyp = word_get_gov_index(word_buffer_get_word_n(config_get_buffer(c), dep));
- if(gov_hyp != gov_ref) return 0;
- }
- }
-#endif
-
-#if 1
for(dep = word_index - 1; (dep >= 0) && (word_get_sent_seg(word_buffer_get_word_n(ref, dep)) == 0); dep--){
gov_ref = word_get_gov_index(word_buffer_get_word_n(ref, dep));
if(gov_ref == word_index){ /* dep is a dependent of word in ref */
@@ -52,9 +31,6 @@ int check_all_dependents_of_word_in_ref_are_in_hyp(config *c, word_buffer *ref,
if(gov_hyp != gov_ref) return 0;
}
}
-
-#endif
-
return 1;
}
@@ -88,33 +64,33 @@ int oracle_parser_arc_eager(config *c, word_buffer *ref, int root_label)
if((s0_label == root_label)
&& check_all_dependents_of_word_in_ref_are_in_hyp(c, ref, s0_index)
){
- return MVT_ROOT;
+ return MVT_PARSER_ROOT;
}
/* word on the top of the stack is an end of sentence marker */
if((word_get_sent_seg(word_buffer_get_word_n(ref, s0_index)) == 1)
&& (word_get_sent_seg(word_buffer_get_word_n(config_get_buffer(c), s0_index)) != 1)){
- return MVT_EOS;
+ return MVT_PARSER_EOS;
}
/* LEFT ARC b0 is the governor and s0 the dependent */
if(s0_gov_index == b0_index){
- return movement_left_code(word_get_label(word_buffer_get_word_n(ref, s0_index)));
+ return movement_parser_left_code(word_get_label(word_buffer_get_word_n(ref, s0_index)));
}
/* RIGHT ARC s0 is the governor and b0 the dependent */
if(b0_gov_index == s0_index){
- return movement_right_code(word_get_label(word_buffer_get_word_n(ref, b0_index)));
+ return movement_parser_right_code(word_get_label(word_buffer_get_word_n(ref, b0_index)));
}
/* REDUCE */
if((stack_nbelem(config_get_stack(c)) > 1)
&& check_all_dependents_of_word_in_ref_are_in_hyp(c, ref, s0_index) /* word on top must have all its dependents */
&& (word_get_gov(stack_top(config_get_stack(c))) != WORD_INVALID_GOV)){ /* word on top of the stack has a governor */
- return MVT_REDUCE;
+ return MVT_PARSER_REDUCE;
}
}
/* SHIFT */
- return MVT_SHIFT;
+ return MVT_PARSER_SHIFT;
}
diff --git a/maca_trans_parser/src/oracle_tagparser_arc_eager.c b/maca_trans_parser/src/oracle_tagparser_arc_eager.c
index d9123e5d72cebff9a8b8a96e3984603ae607c066..c53354f4b012fbadcdc366f35407a71c5eaf6d57 100644
--- a/maca_trans_parser/src/oracle_tagparser_arc_eager.c
+++ b/maca_trans_parser/src/oracle_tagparser_arc_eager.c
@@ -75,9 +75,9 @@ int oracle_tagparser_arc_eager(config *c, word_buffer *ref, int root_label)
/* give a pos to b0 if it does not have one */
if(word_get_pos(b0) == -1){
/* word_set_pos(b0, word_get_pos(word_buffer_get_word_n(ref, b0_index))); */
- /* return movement_tagparse_postag(word_get_pos(b0)); */
+ /* return movement_tagparser_postag(word_get_pos(b0)); */
- return movement_tagparse_postag(word_get_pos(word_buffer_get_word_n(ref, b0_index)));
+ return movement_tagparser_postag_code(word_get_pos(word_buffer_get_word_n(ref, b0_index)));
}
@@ -102,33 +102,33 @@ int oracle_tagparser_arc_eager(config *c, word_buffer *ref, int root_label)
if((s0_label == root_label)
&& check_all_dependents_of_word_in_ref_are_in_hyp(c, ref, s0_index)
){
- return MVT_ROOT;
+ return MVT_TAGPARSER_ROOT;
}
/* word on the top of the stack is an end of sentence marker */
if((word_get_sent_seg(word_buffer_get_word_n(ref, s0_index)) == 1)
&& (word_get_sent_seg(word_buffer_get_word_n(config_get_buffer(c), s0_index)) != 1)){
- return MVT_EOS;
+ return MVT_TAGPARSER_EOS;
}
/* LEFT ARC b0 is the governor and s0 the dependent */
if(s0_gov_index == b0_index){
- return movement_tagparse_left_code(word_get_label(word_buffer_get_word_n(ref, s0_index)));
+ return movement_tagparser_left_code(word_get_label(word_buffer_get_word_n(ref, s0_index)));
}
/* RIGHT ARC s0 is the governor and b0 the dependent */
if(b0_gov_index == s0_index){
- return movement_tagparse_right_code(word_get_label(word_buffer_get_word_n(ref, b0_index)));
+ return movement_tagparser_right_code(word_get_label(word_buffer_get_word_n(ref, b0_index)));
}
/* REDUCE */
if((stack_nbelem(config_get_stack(c)) > 1)
&& check_all_dependents_of_word_in_ref_are_in_hyp(c, ref, s0_index) /* word on top must have all its dependents */
&& (word_get_gov(stack_top(config_get_stack(c))) != WORD_INVALID_GOV)){ /* word on top of the stack has a governor */
- return MVT_REDUCE;
+ return MVT_TAGPARSER_REDUCE;
}
}
/* SHIFT */
- return MVT_SHIFT;
+ return MVT_TAGPARSER_SHIFT;
}
diff --git a/maca_trans_parser/src/simple_decoder_parser_arc_eager.c b/maca_trans_parser/src/simple_decoder_parser_arc_eager.c
index ad65e2eb813ccb2373eb02a902c907f5cffb7205..3a81fc79d7839530580a9e6803a1e070c4dc55b5 100644
--- a/maca_trans_parser/src/simple_decoder_parser_arc_eager.c
+++ b/maca_trans_parser/src/simple_decoder_parser_arc_eager.c
@@ -32,7 +32,6 @@ void print_word_buffer(config *c, dico *dico_labels)
}
}
-
void simple_decoder_parser_arc_eager(context *ctx)
{
FILE *f = (ctx->input_filename)? myfopen(ctx->input_filename, "r") : stdin;
@@ -46,7 +45,7 @@ void simple_decoder_parser_arc_eager(context *ctx)
config *c = NULL;
int result;
float entropy;
- float delta;
+ /* float delta; */
int argmax1, argmax2;
float max1, max2;
int index;
@@ -54,12 +53,12 @@ void simple_decoder_parser_arc_eager(context *ctx)
root_label = dico_string2int(ctx->dico_labels, ctx->root_label);
if(root_label == -1) root_label = 0;
- c = config_initial(f, ctx->mcd_struct, 5);
+ c = config_new(f, ctx->mcd_struct, 5);
while(!config_is_terminal(c)){
config2feat_vec_cff(ctx->features_model, c, ctx->d_perceptron_features, fv, LOOKUP_MODE);
mvt_code = feature_table_argmax(fv, ft, &max);
- mvt_type = movement_type(mvt_code);
- mvt_label = movement_label(mvt_code);
+ mvt_type = movement_parser_type(mvt_code);
+ mvt_label = movement_parser_label(mvt_code);
if(ctx->trace_mode){
index = word_get_index(word_buffer_b0(config_get_buffer(c)));
@@ -68,7 +67,7 @@ void simple_decoder_parser_arc_eager(context *ctx)
stack_print(stdout, c->st);
fprintf(stdout, "\t");
- movement_print(stdout, mvt_code, ctx->dico_labels);
+ movement_parser_print(stdout, mvt_code, ctx->dico_labels);
fprintf(stdout, "\t");
feature_table_argmax_1_2(fv, ft, &argmax1, &max1, &argmax2, &max2);
printf("%f\n", max1 - max2);
@@ -81,9 +80,9 @@ void simple_decoder_parser_arc_eager(context *ctx)
entropy = feature_table_entropy(fv, ft);
/* delta = feature_table_diff_scores(fv, ft); */
feature_table_argmax_1_2(fv, ft, &argmax1, &max1, &argmax2, &max2);
- movement_print(stdout, argmax1, ctx->dico_labels);
+ movement_parser_print(stdout, argmax1, ctx->dico_labels);
printf(":\t%f\n", max1);
- movement_print(stdout, argmax2, ctx->dico_labels);
+ movement_parser_print(stdout, argmax2, ctx->dico_labels);
printf(":\t%f\n", max2);
printf("delta = %f\n", max1 - max2);
@@ -91,34 +90,34 @@ void simple_decoder_parser_arc_eager(context *ctx)
/* printf("entropy = %f delta = %f\n", entropy, delta); */
printf("entropy = %f\n",entropy);
- /* movement_print(stdout, mvt_code, ctx->dico_labels); */
+ /* movement_parser_print(stdout, mvt_code, ctx->dico_labels); */
}
result = 0;
switch(mvt_type){
- case MVT_LEFT :
- result = movement_left_arc(c, mvt_label, max);
+ case MVT_PARSER_LEFT :
+ result = movement_parser_left_arc(c, mvt_label);
break;
- case MVT_RIGHT:
- result = movement_right_arc(c, mvt_label, max);
+ case MVT_PARSER_RIGHT:
+ result = movement_parser_right_arc(c, mvt_label);
break;
- case MVT_REDUCE:
- result = movement_reduce(c, max);
+ case MVT_PARSER_REDUCE:
+ result = movement_parser_reduce(c);
break;
- case MVT_ROOT:
- result = movement_root(c, max, root_label);
+ case MVT_PARSER_ROOT:
+ result = movement_parser_root(c, root_label);
break;
- case MVT_EOS:
- result = movement_eos(c, max);
+ case MVT_PARSER_EOS:
+ result = movement_parser_eos(c);
break;
- case MVT_SHIFT:
- result = movement_shift(c, 1, max);
+ case MVT_PARSER_SHIFT:
+ result = movement_parser_shift(c);
}
if(result == 0){
if(ctx->debug_mode){
fprintf(stdout, "WARNING : movement cannot be executed doing a SHIFT instead !\n");
}
- movement_shift(c, 1, max);
+ movement_parser_shift(c);
}
}
@@ -132,29 +131,3 @@ void simple_decoder_parser_arc_eager(context *ctx)
fclose(f);
}
-#if 0
-void print_word_buffer(config *c, dico *dico_labels)
-{
- int i;
- word *dep;
- char *label;
- int root_position = 0;
-
- for(i=0; i < config_get_buffer(c)->nbelem; i++){
- dep = word_buffer_get_word_n(config_get_buffer(c), i);
- if(word_get_gov(dep) == 0) root_position = i;
- printf("%s\t", word_get_input(dep));
- /* if(word_get_sent_seg(dep) == 1){
- printf("%d\teos\t1\n", root_position - i);
- }
- else{*/
- printf("%d\t", word_get_gov(dep));
- label = (word_get_label(dep) == -1)? NULL : dico_int2string(dico_labels, word_get_label(dep));
- if(label != NULL)
- printf("%s\t0\n", label) ;
- else
- printf("_\t0\n");
- /* } */
- }
-}
-#endif
diff --git a/maca_trans_parser/src/simple_decoder_tagger.c b/maca_trans_parser/src/simple_decoder_tagger.c
index da0e990d72097c336491c378e64644e37809cc4a..bad70b6e5ff3d797f62a3dc436ea1c54a50614ed 100644
--- a/maca_trans_parser/src/simple_decoder_tagger.c
+++ b/maca_trans_parser/src/simple_decoder_tagger.c
@@ -10,18 +10,6 @@
#include"feature_table.h"
#include"dico.h"
-void add_signature_to_words_in_queue(queue *bf, form2pos *f2p)
-{
- int i;
- word *w;
-
- for(i=0; i < queue_nbelem(bf); i++){
- w = queue_elt_n(bf, i);
-
- w->signature = form2pos_get_signature(f2p, w->form);
- }
-}
-
void add_signature_to_words_in_word_buffer(word_buffer *bf, form2pos *f2p)
{
int i;
@@ -34,48 +22,7 @@ void add_signature_to_words_in_word_buffer(word_buffer *bf, form2pos *f2p)
}
}
-void simple_decoder_buffer(context *ctx)
-{
- dico *dico_pos = dico_vec_get_dico(ctx->vocabs, (char *)"POS");
- feature_table *ft = feature_table_load(ctx->perc_model_filename, ctx->verbose);
- int postag;
- feat_vec *fv = feat_vec_new(feature_types_nb);
- float max;
- int i;
- word *w = NULL;
- FILE *f = (ctx->input_filename)? myfopen(ctx->input_filename, "r") : stdin;
- config *c = config_initial(f, ctx->mcd_struct, 0);
-
- /* read a sentence and put it in the buffer */
- while(word_buffer_read_sentence(c->bf)){
- /* queue_remove(c->bf); */
- /* get rid of dummy token */
- if(ctx->f2p)
- add_signature_to_words_in_word_buffer(c->bf, ctx->f2p);
- while(!config_is_terminal(c)){
- config2feat_vec_cff(ctx->features_model, c, ctx->d_perceptron_features, fv, LOOKUP_MODE);
-
- postag = feature_table_argmax(fv, ft, &max);
- if(postag != -1)
- movement_tagger(c, postag, max, 0);
- }
- /* config_print(stdout, c); */
-
- for(i = stack_nbelem(c->st)-1; i >= 0 ; i--){
- w = stack_elt_n(c->st, i);
- printf("%s\t%s\n", w->input, dico_int2string(dico_pos, word_get_pos(w)));
- }
- /* printf("\n"); */
-
- /* config_free(c); */
- c = config_initial(f, ctx->mcd_struct, 0);
- }
- if(ctx->input_filename)
- fclose(f);
-}
-
-
-void simple_decoder_stream(context *ctx)
+void simple_decoder_tagger(context *ctx)
{
config *c;
feat_vec *fv = feat_vec_new(feature_types_nb);
@@ -86,7 +33,7 @@ void simple_decoder_stream(context *ctx)
word *w;
dico *dico_pos = dico_vec_get_dico(ctx->vocabs, (char *)"POS");
int res;
- c = config_initial_no_dummy_word(f, ctx->mcd_struct, 5);
+ c = config_new(f, ctx->mcd_struct, 5);
while(1){
if(ctx->f2p)
@@ -97,10 +44,10 @@ void simple_decoder_stream(context *ctx)
/* feat_vec_print(stdout, fv); */
postag = feature_table_argmax(fv, ft, &max);
/* printf("postag = %d\n", postag); */
-
+
w = word_buffer_b0(c->bf);
printf("%s\t%s\n", w->input, dico_int2string(dico_pos, postag));
-
+
res = movement_tagger(c, postag, max, 1);
/* printf(" current index = %d nb elem = %d\n", c->bf->current_index, c->bf->nbelem); */
@@ -112,12 +59,3 @@ void simple_decoder_stream(context *ctx)
/* config_free(c); */
}
-
-void simple_decoder_tagger(context *ctx)
-{
- if(ctx->stream_mode)
- simple_decoder_stream(ctx);
- else
- simple_decoder_buffer(ctx);
-}
-
diff --git a/maca_trans_parser/src/simple_decoder_tagparser_arc_eager.c b/maca_trans_parser/src/simple_decoder_tagparser_arc_eager.c
index 946c8a4ac0c8a4fcc272b42e8b2d390615d8e150..4b6aac10052fa34c8247075fcd9bdee71e407ec6 100644
--- a/maca_trans_parser/src/simple_decoder_tagparser_arc_eager.c
+++ b/maca_trans_parser/src/simple_decoder_tagparser_arc_eager.c
@@ -22,8 +22,6 @@ void add_signature_to_words_in_word_buffer(word_buffer *bf, form2pos *f2p)
}
}
-
-
void print_word_buffer_tagparser(config *c, dico *dico_labels, dico *dico_pos)
{
int i;
@@ -53,7 +51,6 @@ void print_word_buffer_tagparser(config *c, dico *dico_labels, dico *dico_pos)
}
}
-
void simple_decoder_tagparser_arc_eager(context *ctx)
{
FILE *f = (ctx->input_filename)? myfopen(ctx->input_filename, "r") : stdin;
@@ -67,7 +64,7 @@ void simple_decoder_tagparser_arc_eager(context *ctx)
config *c = NULL;
int result;
float entropy;
- float delta;
+ /* float delta; */
int argmax1, argmax2;
float max1, max2;
int index;
@@ -75,15 +72,15 @@ void simple_decoder_tagparser_arc_eager(context *ctx)
root_label = dico_string2int(ctx->dico_labels, ctx->root_label);
if(root_label == -1) root_label = 0;
- c = config_initial(f, ctx->mcd_struct, 5);
+ c = config_new(f, ctx->mcd_struct, 5);
while(!config_is_terminal(c)){
if(ctx->f2p)
add_signature_to_words_in_word_buffer(c->bf, ctx->f2p);
config2feat_vec_cff(ctx->features_model, c, ctx->d_perceptron_features, fv, LOOKUP_MODE);
mvt_code = feature_table_argmax(fv, ft, &max);
- mvt_type = movement_tagparse_type(mvt_code);
- mvt_label = movement_tagparse_label(mvt_code);
+ mvt_type = movement_tagparser_type(mvt_code);
+ mvt_label = movement_tagparser_label(mvt_code);
if(ctx->trace_mode){
index = word_get_index(word_buffer_b0(config_get_buffer(c)));
@@ -92,7 +89,7 @@ void simple_decoder_tagparser_arc_eager(context *ctx)
stack_print(stdout, c->st);
fprintf(stdout, "\t");
- movement_tagparse_print(stdout, mvt_code, ctx->dico_labels, ctx->dico_postags);
+ movement_tagparser_print(stdout, mvt_code, ctx->dico_labels, ctx->dico_postags);
fprintf(stdout, "\t");
feature_table_argmax_1_2(fv, ft, &argmax1, &max1, &argmax2, &max2);
printf("%f\n", max1 - max2);
@@ -105,9 +102,9 @@ void simple_decoder_tagparser_arc_eager(context *ctx)
entropy = feature_table_entropy(fv, ft);
/* delta = feature_table_diff_scores(fv, ft); */
feature_table_argmax_1_2(fv, ft, &argmax1, &max1, &argmax2, &max2);
- movement_tagparse_print(stdout, argmax1, ctx->dico_labels, ctx->dico_postags);
+ movement_tagparser_print(stdout, argmax1, ctx->dico_labels, ctx->dico_postags);
printf(":\t%f\n", max1);
- movement_tagparse_print(stdout, argmax2, ctx->dico_labels, ctx->dico_postags);
+ movement_tagparser_print(stdout, argmax2, ctx->dico_labels, ctx->dico_postags);
printf(":\t%f\n", max2);
printf("delta = %f\n", max1 - max2);
@@ -115,37 +112,37 @@ void simple_decoder_tagparser_arc_eager(context *ctx)
/* printf("entropy = %f delta = %f\n", entropy, delta); */
printf("entropy = %f\n",entropy);
- /* movement_tagparse_print(stdout, mvt_code, ctx->dico_labels); */
+ /* movement_tagparser_print(stdout, mvt_code, ctx->dico_labels); */
}
result = 0;
switch(mvt_type){
- case MVT_POSTAG :
- result = movement_tagparse_add_pos(c, max, mvt_label);
+ case MVT_TAGPARSER_POSTAG :
+ result = movement_tagparser_add_pos(c, mvt_label);
break;
- case MVT_LEFT :
- result = movement_tagparse_left_arc(c, mvt_label, max);
+ case MVT_TAGPARSER_LEFT :
+ result = movement_tagparser_left_arc(c, mvt_label);
break;
- case MVT_RIGHT:
- result = movement_tagparse_right_arc(c, mvt_label, max);
+ case MVT_TAGPARSER_RIGHT:
+ result = movement_tagparser_right_arc(c, mvt_label);
break;
- case MVT_REDUCE:
- result = movement_tagparse_reduce(c, max);
+ case MVT_TAGPARSER_REDUCE:
+ result = movement_tagparser_reduce(c);
break;
- case MVT_ROOT:
- result = movement_tagparse_root(c, max, root_label);
+ case MVT_TAGPARSER_ROOT:
+ result = movement_tagparser_root(c, root_label);
break;
- case MVT_EOS:
- result = movement_tagparse_eos(c, max);
+ case MVT_TAGPARSER_EOS:
+ result = movement_tagparser_eos(c);
break;
- case MVT_SHIFT:
- result = movement_tagparse_shift(c, 1, max);
+ case MVT_TAGPARSER_SHIFT:
+ result = movement_tagparser_shift(c);
}
if(result == 0){
if(ctx->debug_mode){
fprintf(stdout, "WARNING : movement cannot be executed doing a SHIFT instead !\n");
}
- movement_tagparse_shift(c, 1, max);
+ movement_tagparser_shift(c);
}
}
diff --git a/maca_trans_parser/src/test_mvt_stack.c b/maca_trans_parser/src/test_mvt_stack.c
new file mode 100644
index 0000000000000000000000000000000000000000..711a3a49ffba93a1461c18af5ee6acf36ac63dee
--- /dev/null
+++ b/maca_trans_parser/src/test_mvt_stack.c
@@ -0,0 +1,28 @@
+#include<stdio.h>
+#include<stdlib.h>
+#include"mvt_stack.h"
+#include"util.h"
+
+int main(int argc, char *argv[])
+{
+ mvt_stack *ms = mvt_stack_new();
+ word *w1 = NULL;
+ word *w2 = NULL;
+ int i;
+
+ for(i=0; i < 10; i++){
+ w1 = word_new("1");
+ word_set_index(w1, i + 1);
+ w2 = word_new("1");
+ word_set_index(w2, i + 2);
+ mvt_stack_push(ms, mvt_new(i, w1, w2));
+ }
+ mvt_stack_print(stdout, ms);
+
+ printf("\n");
+
+ while(!mvt_stack_is_empty(ms))
+ mvt_print(stdout, mvt_stack_pop(ms));
+
+
+}