diff --git a/maca_common/include/mcd.h b/maca_common/include/mcd.h
index 5c6747eba43a0cff072a031c0e412ace373d26bb..c45d5d79bb642104b75664d8e368bba0d753f28c 100644
--- a/maca_common/include/mcd.h
+++ b/maca_common/include/mcd.h
@@ -129,4 +129,6 @@ dico_vec *mcd_build_dico_vec(mcd *mcd_struct);
int mcd_wf_code(char *wf);
void mcd_remove_wf_column(mcd *m, int wf_code);
mcd *mcd_copy(mcd *m);
+char *mcd_get_str(mcd *m, int code, int col);
+
#endif
diff --git a/maca_common/include/word.h b/maca_common/include/word.h
index c1eb982862c9c4991cfe902631d96ea9a5ff3fd2..7620b6d8263d46fb1e240f9a72998c9521e7ba8e 100644
--- a/maca_common/include/word.h
+++ b/maca_common/include/word.h
@@ -3,6 +3,8 @@
#include "mcd.h"
+#define WORD_INVALID_GOV 10000
+
#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]
@@ -12,7 +14,7 @@
#define word_get_gov(w) ((w) == NULL) ? 0 : (w)->wf_array[MCD_WF_GOV]
#define word_get_label(w) ((w) == NULL) ? -1 : (w)->wf_array[MCD_WF_LABEL]
#define word_get_stag(w) ((w) == NULL) ? -1 : (w)->wf_array[MCD_WF_STAG]
-#define word_get_sent_seg(w) ((w) == NULL) ? -1 : (w)->wf_array[MCD_WF_SENT_SEG]
+#define word_get_sent_seg(w) ((w) == NULL) ? 0 : (w)->wf_array[MCD_WF_SENT_SEG]
#define word_get_A(w) ((w) == NULL) ? -1 : (w)->wf_array[MCD_WF_A]
#define word_get_B(w) ((w) == NULL) ? -1 : (w)->wf_array[MCD_WF_B]
#define word_get_C(w) ((w) == NULL) ? -1 : (w)->wf_array[MCD_WF_C]
@@ -91,6 +93,7 @@ typedef struct _word {
int label;
char *form;
int index;
+ int is_root;
} word;
word *word_new(char *input);
diff --git a/maca_common/src/mcd.c b/maca_common/src/mcd.c
index 9a2dc6d94d2afefe9b1b70389c6bf71bb844479d..a9313d6babf7eccd421d1fe6d6bd5cb9d688c024 100644
--- a/maca_common/src/mcd.c
+++ b/maca_common/src/mcd.c
@@ -428,3 +428,14 @@ int mcd_wf_code(char *wf)
if(!strcmp(wf, "Z")) return MCD_WF_Z;
return -1;
}
+
+
+char *mcd_get_str(mcd *m, int code, int col)
+{
+ if((col < 0) || (col >= m->nb_col)) return NULL;
+ if(m->representation[col] == MCD_REPRESENTATION_VOCAB)
+ return (m->dico_array[col])? dico_int2string(m->dico_array[col], code) : NULL;
+ return NULL;
+}
+
+
diff --git a/maca_common/src/word.c b/maca_common/src/word.c
index 1a668e4421a96b40af11657254c764c5c894447d..ebf00c6e47670a4988e86be11a1004eebb9f2e0e 100644
--- a/maca_common/src/word.c
+++ b/maca_common/src/word.c
@@ -17,11 +17,12 @@ word *word_new(char *input)
for(i=0; i < MCD_WF_NB; i++) w->wf_array[i] = -1;
- w->wf_array[MCD_WF_GOV] = 0;
+ w->wf_array[MCD_WF_GOV] = WORD_INVALID_GOV;
w->form = NULL;
w->index = -1;
w->signature = -1;
+ w->is_root = 0;
return w;
}
@@ -143,7 +144,7 @@ int word_is_eos(word *w, mcd *mcd_struct)
int word_get_gov_index(word *w)
{
int index;
- if(word_get_gov(w) == 0) return -1;
+ if(word_get_gov(w) == WORD_INVALID_GOV) return -1;
index = (word_get_index(w)) + (word_get_gov(w));
return index;
}
diff --git a/maca_common/src/word_buffer.c b/maca_common/src/word_buffer.c
index 61695c8d52f5cf0a0153c1bc7e8f493f824e8f73..e7cb52b7b6ce4c109f6ccdb2e62bfcf446c0fa2c 100644
--- a/maca_common/src/word_buffer.c
+++ b/maca_common/src/word_buffer.c
@@ -97,15 +97,22 @@ word *word_buffer_get_word_n(word_buffer *wb, int n)
word_buffer *word_buffer_load_mcf(char *mcf_filename, mcd *mcd_struct)
{
- FILE *f = myfopen(mcf_filename, "r");
+ FILE *f;
+
+ if(mcf_filename == NULL)
+ f = stdin;
+ else
+ f = myfopen(mcf_filename, "r");
word_buffer *wb = word_buffer_new(f, mcd_struct, 0);
while(word_buffer_read_next_word(wb) != -1){
/* printf("load word %d\n", wb->nbelem - 1); */
}
- fclose(f);
+ if(mcf_filename != NULL)
+ fclose(f);
return wb;
}
+
int word_buffer_read_next_word(word_buffer *wb)
{
word *w = NULL;
@@ -121,7 +128,7 @@ int word_buffer_move_right(word_buffer *wb)
{
if((wb->nbelem - 1 - wb->current_index) <= wb->lookahead)
word_buffer_read_next_word(wb);
- if(wb->current_index == wb->nbelem - 1) return 0;
+ if(wb->current_index >= wb->nbelem) return 0;
wb->current_index++;
return 1;
}
diff --git a/maca_tools/src/mcf2conll.c b/maca_tools/src/mcf2conll.c
index f518ff5e4b0a8d9efd647ef0e7acef1d9f0b918e..a19adc81dab31d62beab765a0625368d3eb127f9 100644
--- a/maca_tools/src/mcf2conll.c
+++ b/maca_tools/src/mcf2conll.c
@@ -171,36 +171,6 @@ context *context_read_options(int argc, char *argv[])
return ctx;
}
-
-
-
-word_buffer *word_buffer_load_mcf2(char *mcf_filename, mcd *mcd_struct)
-{
- FILE *f;
-
- if(mcf_filename == NULL)
- f = stdin;
- else
- f = myfopen(mcf_filename, "r");
- word_buffer *wb = word_buffer_new(f, mcd_struct, 0);
- while(word_buffer_read_next_word(wb) != -1){
- /* printf("load word %d\n", wb->nbelem - 1); */
- }
- if(mcf_filename != NULL)
- fclose(f);
- return wb;
-}
-
-char *mcd_get_str(mcd *m, int code, int col)
-{
- if((col < 0) || (col >= m->nb_col)) return NULL;
- if(m->representation[col] == MCD_REPRESENTATION_VOCAB)
- return (m->dico_array[col])? dico_int2string(m->dico_array[col], code) : NULL;
- return NULL;
-}
-
-
-
int main(int argc, char *argv[])
{
FILE *output_file;
@@ -208,7 +178,7 @@ int main(int argc, char *argv[])
mcf2conll_check_options(ctx);
- word_buffer *wb = word_buffer_load_mcf2(ctx->mcf_filename, ctx->mcd_struct);
+ word_buffer *wb = word_buffer_load_mcf(ctx->mcf_filename, ctx->mcd_struct);
word *w = NULL;
int form_col = mcd_get_form_col(ctx->mcd_struct);
int pos_col = mcd_get_pos_col(ctx->mcd_struct);
diff --git a/maca_trans_parser/CMakeLists.txt b/maca_trans_parser/CMakeLists.txt
index c5dcf448efae49799baaf604a09a780a70200dad..495f31de62a8ebe90ac4f5c71d9c3f4a878c88ed 100644
--- a/maca_trans_parser/CMakeLists.txt
+++ b/maca_trans_parser/CMakeLists.txt
@@ -4,10 +4,10 @@ set(SOURCES src/context.c
src/movement_tagger.c
src/feat_fct.c
src/global_feat_vec.c
- src/oracle_parser.c
+# src/oracle_parser.c
src/oracle_parser_arc_eager.c
src/oracle_tagger.c
- src/simple_decoder_parser.c
+# src/simple_decoder_parser.c
src/simple_decoder_parser_arc_eager.c
src/simple_decoder_forrest.c
src/simple_decoder_tagger.c
@@ -35,17 +35,17 @@ 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_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_arc_eager_mcf2cff ./src/maca_trans_parser_arc_eager_mcf2cff.c)
target_link_libraries(maca_trans_parser_arc_eager_mcf2cff perceptron)
diff --git a/maca_trans_parser/src/config.c b/maca_trans_parser/src/config.c
index 68e15dd464e277074117c0489c45ca56c34b379c..02c392ecee9fe1d1e3e36c2f849cec6f749e156b 100644
--- a/maca_trans_parser/src/config.c
+++ b/maca_trans_parser/src/config.c
@@ -18,8 +18,6 @@ config *config_new(FILE *f, mcd *mcd_struct, int lookahead)
c->mvt_array_size = 0;
c->mvt_array_nbelem = 0;
c->gfv = global_feat_vec_new();
-
- c->current_index = 1;
return c;
}
@@ -40,22 +38,33 @@ void config_free(config *c)
{
stack_free(c->st);
word_buffer_free(c->bf);
- depset_free(c->ds);
+ /* depset_free(c->ds); */
if(c->mvt_array)free(c->mvt_array);
if(c->gfv)global_feat_vec_free(c->gfv);
free(c);
}
+
+
+
+
int config_is_terminal(config *c)
{
- return word_buffer_end(c->bf);
+ /* 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));
+ /* stack_push(c->st, word_create_dummy(mcd_struct)); */
return c;
}
diff --git a/maca_trans_parser/src/context.c b/maca_trans_parser/src/context.c
index bbdbc3da3a6e0a7608ac35fb4780669a44ecd1e3..be6a90926c29f21b68dcb5c095bd368da94f3e7b 100644
--- a/maca_trans_parser/src/context.c
+++ b/maca_trans_parser/src/context.c
@@ -57,7 +57,7 @@ context *context_new(void)
ctx->f2p_filename = NULL;
ctx->maca_data_path = NULL;
- ctx->language = strdup("fr_stream");
+ ctx->language = strdup("fr");
ctx->root_label = strdup("root");
ctx->d_perceptron_features = NULL;
@@ -159,10 +159,11 @@ context *context_read_options(int argc, char *argv[])
ctx->program_name = strdup(argv[0]);
- static struct option long_options[21] =
+ static struct option long_options[22] =
{
{"help", no_argument, 0, 'h'},
{"verbose", no_argument, 0, 'v'},
+ {"debug", no_argument, 0, 'd'},
{"conll", no_argument, 0, 'c'},
{"stream", no_argument, 0, 'S'},
{"model", required_argument, 0, 'm'},
@@ -187,7 +188,7 @@ context *context_read_options(int argc, char *argv[])
opterr = 0;
- while ((c = getopt_long (argc, argv, "hvcSm:i:n:x:u:r:M:b:f:s:C:F:V:L:D:R:P:", long_options, &option_index)) != -1){
+ while ((c = getopt_long (argc, argv, "hvdcSm:i:n:x:u:r:M:b:f:s:C:F:V:L:D:R:P:", long_options, &option_index)) != -1){
switch (c)
{
case 'h':
@@ -196,6 +197,9 @@ context *context_read_options(int argc, char *argv[])
case 'v':
ctx->verbose = 1;
break;
+ case 'd':
+ ctx->debug_mode = 1;
+ break;
case 'c':
ctx->conll = 1;
break;
diff --git a/maca_trans_parser/src/feat_fct.c b/maca_trans_parser/src/feat_fct.c
index 5740caadc35a99d1b09714354b3179343d3d4fc6..5c665a112a9733853578c851234138d99f0185ac 100644
--- a/maca_trans_parser/src/feat_fct.c
+++ b/maca_trans_parser/src/feat_fct.c
@@ -1,6 +1,7 @@
#include<stdio.h>
#include<stdlib.h>
#include<string.h>
+#include"word_buffer.h"
#include"stack.h"
#include"feat_fct.h"
#include"feat_types.h"
@@ -8,7 +9,7 @@
/* word features */
/* words in the stack */
-int s0g(config *c) {return word_get_gov(stack_s0(config_get_stack(c)));}
+int s0g(config *c) {return (word_get_gov(stack_s0(config_get_stack(c))) == WORD_INVALID_GOV) ? 0 : 1;}
int s0sf(config *c) {return word_get_label(stack_s0(config_get_stack(c)));}
int s0f(config *c) {return word_get_form(stack_s0(config_get_stack(c)));}
@@ -47,7 +48,7 @@ int s0Z(config *c) {return word_get_Z(stack_s0(config_get_stack(c)));}
int s0U1(config *c) {return word_get_U1(stack_s0(config_get_stack(c)));}
int s0sgn(config *c) {return word_get_signature(stack_s0(config_get_stack(c)));}
-int s1g(config *c) {return word_get_gov(stack_s1(config_get_stack(c)));}
+int s1g(config *c) {return (word_get_gov(stack_s1(config_get_stack(c))) == WORD_INVALID_GOV) ? 0 : 1;}
int s1sf(config *c) {return word_get_label(stack_s1(config_get_stack(c)));}
int s1f(config *c) {return word_get_form(stack_s1(config_get_stack(c)));}
@@ -151,6 +152,10 @@ int s3Z(config *c) {return word_get_Z(stack_s3(config_get_stack(c)));}
/* words in the buffer */
+int b0g(config *c) {return (word_get_gov(word_buffer_b0(config_get_buffer(c))) == WORD_INVALID_GOV) ? 0 : 1;}
+int b0sf(config *c) {return word_get_label(word_buffer_b0(config_get_buffer(c)));}
+
+
int b0f(config *c) {return word_get_form(word_buffer_b0(config_get_buffer(c)));}
int b0l(config *c) {return word_get_lemma(word_buffer_b0(config_get_buffer(c)));}
int b0c(config *c) {return word_get_cpos(word_buffer_b0(config_get_buffer(c)));}
@@ -393,165 +398,183 @@ int bm3Z(config *c) {return word_get_Z(word_buffer_bm3(config_get_buffer(c)));}
/* structural features */
-/*int ldep_s0r(config *c){
- word *top = stack_top(config_get_stack(c));
- if(top)
- if(top->last_left_child)
- return top->last_left_child->label;
- return -1;
- }*/
-
int ldep_s0r(config *c){
- word *top = stack_top(c->st);
+ word *gov = stack_s0(config_get_stack(c));
int i;
-
- if(top){
- if(word_get_index(top) >= c->ds->length) return -1;
- for(i=word_get_index(top); i > 0; i--)
- if(c->ds->array[i].gov == top)
- return i;
+ word *dep;
+ int dist;
+
+ if(gov){
+ for(i=word_get_index(gov) - 1; i > 0 ; i--){
+ dep = word_buffer_get_word_n(config_get_buffer(c), i);
+ if(word_get_sent_seg(dep) == 1) return -1;
+ dist = word_get_index(gov) - i;
+ if(word_get_gov(dep) == dist){
+ return word_get_label(dep);
+ }
+ }
}
return -1;
}
-/*int rdep_s0r(config *c){
- word *top = stack_top(c->st);
-
- if(top)
- if(top->last_right_child)
- return top->last_right_child->label;
- return -1;
- }*/
-
-int rdep_s0r(config *c){
- word *top = stack_top(c->st);
+int ldep_s1r(config *c){
+ word *gov = stack_s1(config_get_stack(c));
int i;
-
- if(top)
- for(i=word_get_index(top); i < c->ds->length; i++)
- if(c->ds->array[i].gov == top)
- return i;
+ word *dep;
+ int dist;
+
+ if(gov){
+ for(i=word_get_index(gov) - 1; i > 0 ; i--){
+ dep = word_buffer_get_word_n(config_get_buffer(c), i);
+ if(word_get_sent_seg(dep) == 1) return -1;
+ dist = word_get_index(gov) - i;
+ if(word_get_gov(dep) == dist){
+ return word_get_label(dep);
+ }
+ }
+ }
return -1;
}
-/*int ldep_b0r(config *c){
- word *top = queue_elt_n(c->bf,0);
+int ldep_b0r(config *c){
+ word *gov = word_buffer_b0(config_get_buffer(c));
+ int i;
+ word *dep;
+ int dist;
- if(top)
- if(top->last_left_child)
- return top->last_left_child->label;
+ if(gov){
+ for(i=word_get_index(gov) - 1; i > 0 ; i--){
+ dep = word_buffer_get_word_n(config_get_buffer(c), i);
+ if(word_get_sent_seg(dep) == 1) return -1;
+ dist = word_get_index(gov) - i;
+ if(word_get_gov(dep) == dist){
+ return word_get_label(dep);
+ }
+ }
+ }
return -1;
- }*/
+}
-int ldep_b0r(config *c){
- word *top = word_buffer_b0(c->bf);
+int rdep_s0r(config *c){
+ word *gov = stack_s0(config_get_stack(c));
int i;
-
- if(top){
- if(word_get_index(top) >= c->ds->length) return -1;
- for(i=word_get_index(top); i > 0; i--)
- if(c->ds->array[i].gov == top)
- return i;
+ word *dep;
+ int dist;
+
+ if(gov){
+ for(i=word_get_index(gov) + 1; i < word_buffer_get_nbelem(config_get_buffer(c)) ; i++){
+ dep = word_buffer_get_word_n(config_get_buffer(c), i);
+ if(i >= word_get_index(word_buffer_b0(config_get_buffer(c)))) return -1;
+ dist = i - word_get_index(gov);
+ if(word_get_gov(dep) == - dist){
+ return word_get_label(dep);
+ }
+ }
}
-
return -1;
}
-
-
-
-/*int rdep_b0r(config *c){
- word *top = queue_elt_n(c->bf,0);
+int rdep_s1r(config *c){
+ word *gov = stack_s1(config_get_stack(c));
+ int i;
+ word *dep;
+ int dist;
- if(top)
- if(top->last_right_child)
- return top->last_right_child->label;
+ if(gov){
+ for(i=word_get_index(gov) + 1; i < word_buffer_get_nbelem(config_get_buffer(c)) ; i++){
+ dep = word_buffer_get_word_n(config_get_buffer(c), i);
+ if(i >= word_get_index(word_buffer_b0(config_get_buffer(c)))) return -1;
+ dist = i - word_get_index(gov);
+ if(word_get_gov(dep) == - dist){
+ return word_get_label(dep);
+ }
+ }
+ }
return -1;
- }*/
+}
int rdep_b0r(config *c){
- word *top = word_buffer_b0(c->bf);
+ word *gov = word_buffer_b0(config_get_buffer(c));
int i;
-
- if(top)
- for(i=word_get_index(top); i < c->ds->length; i++)
- if(c->ds->array[i].gov == top)
- return i;
+ word *dep;
+ int dist;
+
+ if(gov){
+ for(i=word_get_index(gov) + 1; i < word_buffer_get_nbelem(config_get_buffer(c)) ; i++){
+ dep = word_buffer_get_word_n(config_get_buffer(c), i);
+ if(i >= word_get_index(word_buffer_b0(config_get_buffer(c)))) return -1;
+ dist = i - word_get_index(gov);
+ if(word_get_gov(dep) == - dist){
+ return word_get_label(dep);
+ }
+ }
+ }
return -1;
}
-
-
-/*int ndep_b0(config *c){
- word *top = queue_elt_n(c->bf,0);
- int nb = 0;
- word *child;
-
- if(top){
- for(child = top->first_left_child; child; child = child->next_sibling)
- nb++;
-
- for(child = top->first_right_child; child; child = child->next_sibling)
- nb++;
- }
- return (nb > 6)? 6 : nb;
- }*/
-
-
int ndep_b0(config *c){
- word *top = word_buffer_b0(c->bf);
- int i = 0;
+ word *gov = word_buffer_b0(config_get_buffer(c));
+ int i;
int n = 0;
-
- for(i=0; i < c->ds->length; i++)
- if(c->ds->array[i].gov == top)
- n++;
-
+ word *dep;
+ int dist;
+
+ if(gov){
+ for(i=word_get_index(gov) - 1; i > 0 ; i--){
+ dep = word_buffer_get_word_n(config_get_buffer(c), i);
+ if(word_get_sent_seg(dep) == 1) break;
+ dist = word_get_index(gov) - i;
+ if(word_get_gov(dep) == dist)
+ n++;
+ }
+ for(i=word_get_index(gov) + 1; i < word_buffer_get_nbelem(config_get_buffer(c)) ; i++){
+ dep = word_buffer_get_word_n(config_get_buffer(c), i);
+ if(i >= word_get_index(word_buffer_b0(config_get_buffer(c)))) break;
+ dist = i - word_get_index(gov);
+ if(word_get_gov(dep) == - dist)
+ n++;
+ }
+ }
return (n > 6)? 6 : n;
}
-
-/*
int ndep_s0(config *c){
- word *top = stack_top(c->st);
- int nb = 0;
- word *child;
-
- if(top){
- for(child = top->first_left_child; child; child = child->next_sibling)
- nb++;
-
- for(child = top->first_right_child; child; child = child->next_sibling)
- nb++;
- }
- return (nb > 6)? 6 : nb;
- }*/
-
-int ndep_s0(config *c){
- word *top = stack_top(c->st);
- int i = 0;
+ word *gov = stack_s0(config_get_stack(c));
+ int i;
int n = 0;
-
- for(i=0; i < c->ds->length; i++)
- if(c->ds->array[i].gov == top)
- n++;
-
+ word *dep;
+ int dist;
+
+ if(gov){
+ for(i=word_get_index(gov) - 1; i > 0 ; i--){
+ dep = word_buffer_get_word_n(config_get_buffer(c), i);
+ if(word_get_sent_seg(dep) == 1) break;
+ dist = word_get_index(gov) - i;
+ if(word_get_gov(dep) == dist)
+ n++;
+ }
+ for(i=word_get_index(gov) + 1; i < word_buffer_get_nbelem(config_get_buffer(c)) ; i++){
+ dep = word_buffer_get_word_n(config_get_buffer(c), i);
+ if(i >= word_get_index(word_buffer_b0(config_get_buffer(c)))) break;
+ dist = i - word_get_index(gov);
+ if(word_get_gov(dep) == - dist)
+ n++;
+ }
+ }
return (n > 6)? 6 : n;
}
-
-
/* distance features */
int dist_s0_b0(config *c){
int dist;
- if(stack_is_empty(c->st) || word_buffer_is_empty(c->bf))
+ if(stack_is_empty(config_get_stack(c)) || word_buffer_is_empty(config_get_buffer(c)))
return 0;
- dist = word_get_index(word_buffer_b0(c->bf)) - word_get_index(stack_top(c->st));
+ dist = word_get_index(word_buffer_b0(config_get_buffer(c))) - word_get_index(stack_top(config_get_stack(c)));
return (abs(dist) > 6)? 6 : dist;
}
@@ -560,12 +583,12 @@ int dist_s0_b0(config *c){
int sh(config *c) /* stack height */
{
- return (c->st->size > 7)? 7 : c->st->size;
+ return (config_get_stack(c)->size > 7)? 7 : config_get_stack(c)->size;
}
int bh(config *c) /* buffer size */
{
- return (c->bf->size > 7)? 7 : c->bf->size;
+ return (config_get_buffer(c)->size > 7)? 7 : config_get_buffer(c)->size;
}
int dh(config *c) /* depset size */
diff --git a/maca_trans_parser/src/feat_fct.h b/maca_trans_parser/src/feat_fct.h
index 6f91a3afc8d1e91aae837918ec2fee505f3dbc80..41af327a4465b5cf6c19b97ab82d23bbbe44c261 100644
--- a/maca_trans_parser/src/feat_fct.h
+++ b/maca_trans_parser/src/feat_fct.h
@@ -155,6 +155,9 @@ int s3Y(config *c);
int s3Z(config *c);
int s3r(config *c);
+int b0g(config *c);
+int b0sf(config *c);
+
int b0f(config *c);
int b0l(config *c);
@@ -413,6 +416,8 @@ int gs0p(config *c);
int ldep_s0r(config *c);
int rdep_s0r(config *c);
+int ldep_s1r(config *c);
+int rdep_s1r(config *c);
int ndep_b0(config *c);
int ndep_s0(config *c);
int ldep_b0r(config *c);
diff --git a/maca_trans_parser/src/feat_lib.c b/maca_trans_parser/src/feat_lib.c
index 6a2293f5327804028f34d66605af322ffaadeb14..a60a51103fb1f2ce52f15d9af69a291e18cd567e 100644
--- a/maca_trans_parser/src/feat_lib.c
+++ b/maca_trans_parser/src/feat_lib.c
@@ -177,6 +177,9 @@ feat_lib *feat_lib_build(void)
feat_lib_add(fl, FEAT_TYPE_Y, (char *)"s3Y", s3Y);
feat_lib_add(fl, FEAT_TYPE_Y, (char *)"s3Z", s3Z);
+ feat_lib_add(fl, FEAT_TYPE_FORM, (char *)"b0g", b0g);
+ feat_lib_add(fl, FEAT_TYPE_FORM, (char *)"b0sf", b0sf);
+
feat_lib_add(fl, FEAT_TYPE_FORM, (char *)"b0f", b0f);
feat_lib_add(fl, FEAT_TYPE_LEMMA, (char *)"b0l", b0l);
feat_lib_add(fl, FEAT_TYPE_CPOS, (char *)"b0c", b0c);
@@ -430,6 +433,8 @@ feat_lib *feat_lib_build(void)
feat_lib_add(fl, FEAT_TYPE_LABEL, (char *)"ldep_s0r", ldep_s0r);
feat_lib_add(fl, FEAT_TYPE_LABEL, (char *)"rdep_s0r", rdep_s0r);
+ feat_lib_add(fl, FEAT_TYPE_LABEL, (char *)"ldep_s1r", ldep_s1r);
+ feat_lib_add(fl, FEAT_TYPE_LABEL, (char *)"rdep_s1r", rdep_s1r);
feat_lib_add(fl, FEAT_TYPE_LABEL, (char *)"ldep_b0r", ldep_b0r);
feat_lib_add(fl, FEAT_TYPE_LABEL, (char *)"rdep_b0r", rdep_b0r);
diff --git a/maca_trans_parser/src/maca_trans_parser.c b/maca_trans_parser/src/maca_trans_parser.c
index f65c10dd4ee3ac23498ae6204d34bec9a4d8e427..43a304507aca2b817b028eeb5b4114b7e11e1c3f 100644
--- a/maca_trans_parser/src/maca_trans_parser.c
+++ b/maca_trans_parser/src/maca_trans_parser.c
@@ -11,7 +11,7 @@
#include"feature_table.h"
#include"dico.h"
#include"beam.h"
-#include"simple_decoder_parser.h"
+#include"simple_decoder_parser_arc_eager.h"
/*#include"dnn_decoder.h"*/
#include"config2feat_vec.h"
@@ -112,14 +112,14 @@ int main(int argc, char *argv[])
return 1;
}
- ctx->mvt_nb = ctx->dico_labels->nbelem * 2 + 1;
+ 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(ctx);
+ 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);
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 ae143cd277cd38eadedbd5aa2086de5b31c073a3..ea9d8cf9fded0975ed3d27969a0164cf56dfb680 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
@@ -53,7 +53,6 @@ void generate_training_file_stream(FILE *output_file, context *ctx)
int root_label = dico_string2int(ctx->dico_labels, (char *) ctx->root_label);
word_buffer *ref = word_buffer_load_mcf(ctx->input_filename, ctx->mcd_struct);
FILE *mcf_file = myfopen(ctx->input_filename, "r");
- int start_sentence_index = 0;
/* create an mcd that corresponds to ctx->mcd_struct, but without gov and label */
/* the idea is to ignore syntax in the mcf file that will be read */
@@ -62,8 +61,9 @@ void generate_training_file_stream(FILE *output_file, context *ctx)
mcd *mcd_struct_hyp = mcd_copy(ctx->mcd_struct);
mcd_remove_wf_column(mcd_struct_hyp, MCD_WF_GOV);
mcd_remove_wf_column(mcd_struct_hyp, MCD_WF_LABEL);
+ mcd_remove_wf_column(mcd_struct_hyp, MCD_WF_SENT_SEG);
- c = config_initial_no_dummy_word(mcf_file, mcd_struct_hyp, 5);
+ c = config_initial(mcf_file, mcd_struct_hyp, 5);
while(!word_buffer_end(ref) && (sentence_nb < ctx->sent_nb)){
/*printf("************ REF ************\n");
@@ -72,23 +72,20 @@ 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_parser_arc_eager(c, ref, start_sentence_index, root_label);
+ mvt_code = oracle_parser_arc_eager(c, ref, root_label);
mvt_type = movement_type(mvt_code);
mvt_label = movement_label(mvt_code);
-
- /* config_print(stdout,c); */
- /* movement_print(stdout, mvt_code, ctx->dico_labels); */
-
+
+ if(ctx->debug_mode){
+ config_print(stdout,c);
+ movement_print(stdout, mvt_code, ctx->dico_labels);
+ }
fprintf(output_file, "%d", mvt_code);
feat_vec_print(output_file, fv);
if(mvt_type == MVT_EOS){
movement_eos(c, 0);
sentence_nb++;
- start_sentence_index = word_get_index(word_buffer_b0(config_get_buffer(c))) - 1;
-
-
- /* config_print(stdout,c); */
if(word_buffer_is_last(ref))
break;
}
@@ -147,7 +144,7 @@ int main(int argc, char *argv[])
fprintf(stderr, "cannot find label names\n");
return 1;
}
- ctx->mvt_nb = ctx->dico_labels->nbelem * 2 + 1;
+ ctx->mvt_nb = ctx->dico_labels->nbelem * 2 + 3;
feat_model_compute_ranges(ctx->features_model, ctx->mcd_struct, ctx->mvt_nb);
diff --git a/maca_trans_parser/src/movement_parser_arc_eager.c b/maca_trans_parser/src/movement_parser_arc_eager.c
index 3941705e5e0ab38c70aac18847384dec3060f1e8..564da2a0e7c849eb3e6710286782b888a4e6a29a 100644
--- a/maca_trans_parser/src/movement_parser_arc_eager.c
+++ b/maca_trans_parser/src/movement_parser_arc_eager.c
@@ -20,20 +20,20 @@ void movement_print(FILE *f, int mvt_code, dico *dico_labels){
int movement_type(int mvt)
{
- if(mvt == MVT_SHIFT) return MVT_SHIFT; /* 0 is the code of shift */
- if(mvt == MVT_REDUCE) return MVT_REDUCE; /* 1 is the code of reduce */
- if(mvt == MVT_ROOT) return MVT_ROOT; /* 2 is the code of root */
- if(mvt == MVT_EOS) return MVT_EOS; /* 3 is the code of root */
- if(mvt % 2 == 0) return MVT_LEFT; /* even movements are left movements */
- return MVT_RIGHT; /* odd movements are right movements */
+ 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 */
}
int movement_label(int mvt)
{
- if(mvt == MVT_SHIFT) return -1; /* 0 is the code of shift */
- if(mvt == MVT_REDUCE) return -1; /* 1 is the code of reduce */
- if(mvt == MVT_ROOT) return -1; /* 2 is the code of root */
- if(mvt == MVT_EOS) return -1; /* 3 is the code of eos */
+ 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 % 2 == 0) /* even codes correspond to left movements */
return mvt / 2 - 2;
return (mvt - 1) / 2 - 2; /* odd codes correspond to right movements */
@@ -41,16 +41,10 @@ int movement_label(int mvt)
int movement_eos(config *c, float score)
{
- if(stack_is_empty(config_get_stack(c))) return 1;
+ 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);
-
- /* perform all pending reduce */
- while(movement_reduce(c,0));
-
- /* remove root from stack */
- stack_pop(config_get_stack(c));
+ word_set_sent_seg(stack_top(config_get_stack(c)), 1);
config_add_mvt(c, MVT_EOS);
return 1;
@@ -58,18 +52,16 @@ int movement_eos(config *c, float score)
int movement_left_arc(config *c, int label, float score)
{
- /* the dummy word cannot be a dependent */
- /* if(stack_height(config_get_stack(c)) < 2) return 0; */
+ 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))) != 0) return 0;
+ 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));
- /* printf("create left arc %d <- %d dist = %d\n", word_get_index(dep), word_get_index(gov), dist); */
-
/* create a new dependency */
word_set_gov(dep, dist);
word_set_label(dep, label);
@@ -82,13 +74,10 @@ int movement_left_arc(config *c, int label, float score)
int movement_right_arc(config *c, int label, float score)
{
if(stack_is_empty(config_get_stack(c))) return 0;
- /* if(word_buffer_is_empty(config_get_buffer(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));
-
- /* printf("create right arc %d -> %d dist = %d\n", word_get_index(gov), word_get_index(dep), dist); */
/* create a new dependency */
word_set_gov(dep, dist);
@@ -114,7 +103,7 @@ int movement_reduce(config *c, float score)
{
if(stack_is_empty(config_get_stack(c))) return 0;
/* word on top of stack must have a governor */
- if(word_get_gov(stack_top(config_get_stack(c))) == 0) return 0;
+ 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;
@@ -122,10 +111,12 @@ int movement_reduce(config *c, float score)
int movement_root(config *c, float score, int root_code)
{
- word *b0 = word_buffer_b0(config_get_buffer(c));
- word_set_gov(b0, 0);
- word_set_label(b0, 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_add_mvt(c, MVT_ROOT);
return 1;
}
-
diff --git a/maca_trans_parser/src/oracle_parser_arc_eager.c b/maca_trans_parser/src/oracle_parser_arc_eager.c
index 2e41c609187f4ef488478728bb212ab78565aee0..e77d9e4cb96eab8f079f458cff12641f620c92fa 100644
--- a/maca_trans_parser/src/oracle_parser_arc_eager.c
+++ b/maca_trans_parser/src/oracle_parser_arc_eager.c
@@ -4,65 +4,104 @@
#include"word_buffer.h"
#include"movement_parser_arc_eager.h"
-int check_all_dependents_of_word_in_ref_are_in_hyp(config *c, word_buffer *ref, int word_index, int start_sentence_index)
+int check_all_dependents_of_word_in_ref_are_in_hyp(config *c, word_buffer *ref, int word_index)
{
int dep;
int gov_ref;
int gov_hyp;
- int max_sent_length = 300;
- int max = ((start_sentence_index + max_sent_length) > ref->nbelem)? ref->nbelem : (start_sentence_index + max_sent_length);
- for(dep=start_sentence_index; dep < max; dep++){
+ int sentence_change;
+
+#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){ /* found a dependent of word in ref */
+ /* look for a dependency in hyp such that its dependent is dep */
+ 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){ /* found a dependent of word in ref */
/* look for a dependency in hyp such that its dependent is dep */
- /* printf("found a dep of word %d in ref, it is %d\n", word_index, dep); */
gov_hyp = word_get_gov_index(word_buffer_get_word_n(config_get_buffer(c), dep));
- /* printf("gov of %d in hyp is %d\n", dep,gov_hyp); */
if(gov_hyp != gov_ref) return 0;
- /*
- if((dep >= c->ds->length)
- || (c->ds->array[dep].gov == NULL)
- || (word_get_index(c->ds->array[dep].gov) != word_index)
- || (c->ds->array[dep].label != word_get_label(ref->words[dep])))
- return 0;*/
}
}
+#endif
+
+#if 0
+ sentence_change = 0;
+ for(dep = word_index - 1; (dep > 0) && (word_get_sent_seg(word_buffer_get_word_n(ref, dep)) == 0); dep--){
+ /* printf("dep = %d\n", dep); */
+ if(word_get_sent_seg(word_buffer_get_word_n(ref, dep)) == 1)
+ sentence_change = 1;
+ gov_ref = word_get_gov_index(word_buffer_get_word_n(ref, dep));
+ if(gov_ref == word_index){ /* found a dependent of word in ref */
+ /* look for a dependency in hyp such that its dependent is dep */
+ gov_hyp = word_get_gov_index(word_buffer_get_word_n(config_get_buffer(c), dep));
+ if(gov_hyp != gov_ref) return 0;
+ }
+ }
+
+ sentence_change = 0;
+ for(dep = word_index + 1;
+ (dep < word_buffer_get_nbelem(ref)) && (sentence_change == 0); dep++){
+ /* printf("dep = %d\n", dep); */
+ if(word_get_sent_seg(word_buffer_get_word_n(ref, dep)) == 1)
+ sentence_change = 1;
+ gov_ref = word_get_gov_index(word_buffer_get_word_n(ref, dep));
+ if(gov_ref == word_index){ /* found a dependent of word in ref */
+ /* look for a dependency in hyp such that its dependent is dep */
+ gov_hyp = word_get_gov_index(word_buffer_get_word_n(config_get_buffer(c), dep));
+ if(gov_hyp != gov_ref) return 0;
+ }
+ }
+
+#endif
+
return 1;
}
-int oracle_parser_arc_eager(config *c, word_buffer *ref, int start_sentence_index, int root_label)
+int oracle_parser_arc_eager(config *c, word_buffer *ref, int root_label)
{
word *s0; /* word on top of stack */
word *b0; /* next word in the bufer */
int s0_index, b0_index;
int s0_gov_index, b0_gov_index;
- int b0_label;
- int b0_label_in_hyp;
+ int s0_label;
+ /* int s0_label_in_hyp; */
- b0 = word_buffer_b0(config_get_buffer(c));
- b0_index = word_get_index(b0);
- b0_gov_index = word_get_gov_index(word_buffer_get_word_n(ref, b0_index));
- b0_label = word_get_label(word_buffer_get_word_n(ref, b0_index));
- b0_label_in_hyp = word_get_label(word_buffer_get_word_n(config_get_buffer(c), b0_index));
-
- /* b0 is the root of the sentence */
- if((b0_label == root_label) && (b0_label_in_hyp != root_label)){
- return MVT_ROOT;
- }
-
/* if(!stack_is_empty(config_get_stack(c)) && !word_buffer_is_empty(config_get_buffer(c))){ */
if(!stack_is_empty(config_get_stack(c))){
+
+ b0 = word_buffer_b0(config_get_buffer(c));
+ b0_index = word_get_index(b0);
+ b0_gov_index = word_get_gov_index(word_buffer_get_word_n(ref, b0_index));
+
s0 = stack_top(config_get_stack(c));
s0_index = word_get_index(s0);
s0_gov_index = word_get_gov_index(word_buffer_get_word_n(ref, s0_index));
+ s0_label = word_get_label(word_buffer_get_word_n(ref, s0_index));
+ /* s0_label_in_hyp = word_get_label(word_buffer_get_word_n(config_get_buffer(c), s0_index)); */
/* printf("s0_index = %d b0_index = %d\n", s0_index, b0_index);
printf("dans ref gov de s0 (%d) = %d\n", s0_index, s0_gov_index);
printf("dans ref gov de b0 (%d) = %d\n", b0_index, b0_gov_index);*/
+ /* s0 is the root of the sentence */
+ if((s0_label == root_label)
+ && check_all_dependents_of_word_in_ref_are_in_hyp(c, ref, s0_index)
+ ){
+ return MVT_ROOT;
+ }
+
/* word on the top of the stack is an end of sentence marker */
- if(word_get_sent_seg(s0) == 1){
+ 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;
}
@@ -76,13 +115,11 @@ int oracle_parser_arc_eager(config *c, word_buffer *ref, int start_sentence_inde
return movement_right_code(word_get_label(word_buffer_get_word_n(ref, b0_index)));
}
/* REDUCE */
- if(
- /* (stack_height(config_get_stack(c)) > 2) */
- check_all_dependents_of_word_in_ref_are_in_hyp(c, ref, s0_index, start_sentence_index)
- && (word_get_gov(stack_top(config_get_stack(c))) != 0)) /* word on top of the stack has a goveror */
- {
- return MVT_REDUCE;
- }
+ if((stack_height(config_get_stack(c)) > 2)
+ && check_all_dependents_of_word_in_ref_are_in_hyp(c, ref, s0_index)
+ && (word_get_gov(stack_top(config_get_stack(c))) != WORD_INVALID_GOV)){ /* word on top of the stack has a goveror */
+ return MVT_REDUCE;
+ }
}
/* SHIFT */
diff --git a/maca_trans_parser/src/oracle_parser_arc_eager.h b/maca_trans_parser/src/oracle_parser_arc_eager.h
index 9b92d5d284f7e183c726efc1042af06f2acbf734..af3131214143e88c3fb267f538ab5cbb9ba3f59f 100644
--- a/maca_trans_parser/src/oracle_parser_arc_eager.h
+++ b/maca_trans_parser/src/oracle_parser_arc_eager.h
@@ -6,6 +6,6 @@
#include"word_buffer.h"
-int oracle_parser_arc_eager(config *c, word_buffer *ref, int start_sentence_index, int root_label);
+int oracle_parser_arc_eager(config *c, word_buffer *ref, int root_label);
#endif
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 1be8bbf27129f9e0d3e51352888fcaadb81d2429..03b1f7c8863017810eda9edfc1b5872882a6b2d8 100644
--- a/maca_trans_parser/src/simple_decoder_parser_arc_eager.c
+++ b/maca_trans_parser/src/simple_decoder_parser_arc_eager.c
@@ -15,19 +15,23 @@ 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));
- 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\t", label) ;
- else
- printf("_\t");
- if((label != NULL) && !strcmp(label, "eos"))
- printf("1\n");
- else
- printf("0\n");
+ 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");
+ }
}
}
@@ -43,65 +47,55 @@ void simple_decoder_parser_arc_eager(context *ctx)
float max;
feat_vec *fv = feat_vec_new(feature_types_nb);
config *c = NULL;
+ int result;
root_label = dico_string2int(ctx->dico_labels, ctx->root_label);
if(root_label == -1) root_label = 0;
- c = config_initial_no_dummy_word(f, ctx->mcd_struct, 5);
- while(1){
+ c = config_initial(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);
- config_print(stdout, c);
- movement_print(stdout, mvt_code, ctx->dico_labels);
+ if(ctx->debug_mode){
+ fprintf(stdout, "***********************************\n");
+ config_print(stdout, c);
+ movement_print(stdout, mvt_code, ctx->dico_labels);
+ }
+ result = 0;
+ switch(mvt_type){
+ case MVT_LEFT :
+ result = movement_left_arc(c, mvt_label, max);
+ break;
+ case MVT_RIGHT:
+ result = movement_right_arc(c, mvt_label, max);
+ break;
+ case MVT_REDUCE:
+ result = movement_reduce(c, max);
+ break;
+ case MVT_ROOT:
+ result = movement_root(c, max, root_label);
+ break;
+ case MVT_EOS:
+ result = movement_eos(c, max);
+ break;
+ case MVT_SHIFT:
+ result = movement_shift(c, 1, max);
+ }
- if(mvt_type == MVT_LEFT)
- if(movement_left_arc(c, mvt_label, max)){
- if(word_buffer_is_last(config_get_buffer(c)))
- break;
- continue;
+ if(result == 0){
+ if(ctx->debug_mode){
+ fprintf(stdout, "WARNING : movement cannot be executed !\n");
}
-
- if(mvt_type == MVT_RIGHT)
- if(movement_right_arc(c, mvt_label, max)){
- if(word_buffer_is_last(config_get_buffer(c)))
- break;
- continue;
- }
-
- if(mvt_type == MVT_REDUCE)
- if(movement_reduce(c, max)){
- if(word_buffer_is_last(config_get_buffer(c)))
- break;
- continue;
- }
-
- if(mvt_type == MVT_ROOT)
- if(movement_root(c, max, root_label)){
- if(word_buffer_is_last(config_get_buffer(c)))
- break;
- continue;
- }
-
- /* if(mvt_type == MVT_EOS)
- if(movement_eos(c, max)){
- if(word_buffer_is_last(config_get_buffer(c)))
- break;
- continue;
- }
- */
- if(word_buffer_is_last(config_get_buffer(c)))
- break;
-
- movement_shift(c, 1, max);
-
+ movement_shift(c, 1, max);
+ }
}
print_word_buffer(c, ctx->dico_labels);
- /* config_free(c); */
+ config_free(c);
feat_vec_free(fv);
feature_table_free(ft);
if(ctx->input_filename)
diff --git a/maca_trans_parser/src/simple_decoder_tagger.c b/maca_trans_parser/src/simple_decoder_tagger.c
index 9e3341c362a39892648f55cbf32b665856cdff48..da0e990d72097c336491c378e64644e37809cc4a 100644
--- a/maca_trans_parser/src/simple_decoder_tagger.c
+++ b/maca_trans_parser/src/simple_decoder_tagger.c
@@ -44,7 +44,7 @@ void simple_decoder_buffer(context *ctx)
int i;
word *w = NULL;
FILE *f = (ctx->input_filename)? myfopen(ctx->input_filename, "r") : stdin;
- config *c = config_initial_no_dummy_word(f, ctx->mcd_struct, 0);
+ 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)){
@@ -68,7 +68,7 @@ void simple_decoder_buffer(context *ctx)
/* printf("\n"); */
/* config_free(c); */
- c = config_initial_no_dummy_word(f, ctx->mcd_struct, 0);
+ c = config_initial(f, ctx->mcd_struct, 0);
}
if(ctx->input_filename)
fclose(f);