diff --git a/maca_common/include/word_buffer.h b/maca_common/include/word_buffer.h
index a2e358b2cae411be0b609b734f4f28959967d690..f24ab4cee929fe99b93529b94032c1c9ddf25cfa 100644
--- a/maca_common/include/word_buffer.h
+++ b/maca_common/include/word_buffer.h
@@ -44,6 +44,7 @@ int word_buffer_read_next_word(word_buffer *wb);
int word_buffer_move_right(word_buffer *wb);
int word_buffer_move_left(word_buffer *wb);
void word_buffer_print(FILE *f, word_buffer *wb);
+void word_buffer_print_compact(FILE *f, word_buffer *wb);
int word_buffer_is_empty(word_buffer *wb);
int word_buffer_is_last(word_buffer *wb);
int word_buffer_end(word_buffer *wb);
diff --git a/maca_common/src/mcd.c b/maca_common/src/mcd.c
index 5f068e21995509830abfc1c1fb5a2bf25279012f..64be8859569b5af09667538c9a46239ad8d7511c 100644
--- a/maca_common/src/mcd.c
+++ b/maca_common/src/mcd.c
@@ -56,7 +56,7 @@ mcd *mcd_copy(mcd *m)
mcd *copy = mcd_new(m->nb_col);
for(i=0; i < MCD_WF_NB; i++)
- copy->wf2col[i] = m->wf2col[i] = -1;
+ copy->wf2col[i] = m->wf2col[i];
for(i=0; i < m->nb_col; i++){
copy->representation[i] = m->representation[i];
diff --git a/maca_common/src/word.c b/maca_common/src/word.c
index d4b25c03d074649e499d1c6aa6cf19eb632c4f3e..be7d2f8d26f9763fb82cc89df9fd9412a79d89f6 100644
--- a/maca_common/src/word.c
+++ b/maca_common/src/word.c
@@ -145,8 +145,8 @@ int word_get_gov_relative_index(word *w)
int index;
if(word_get_gov(w) == 0) return -1;
- printf("in word_get_gov_rel_index(%d)\n", word_get_relative_index(w));
- printf("gov = %d\n", word_get_gov(w));
+ /* printf("in word_get_gov_rel_index(%d)\n", word_get_relative_index(w));
+ printf("gov = %d\n", word_get_gov(w)); */
index = (word_get_relative_index(w)) + (word_get_gov(w));
/* printf("index = %d\n", index); */
diff --git a/maca_common/src/word_buffer.c b/maca_common/src/word_buffer.c
index 06d6b8d1e5fec585ed4a7639cdedc6551fea0f53..aabe1876898902cf68621439b6b14c3ef0173a31 100644
--- a/maca_common/src/word_buffer.c
+++ b/maca_common/src/word_buffer.c
@@ -45,6 +45,26 @@ void word_buffer_print(FILE *f, word_buffer *wb)
if(w){ fprintf(f, "[ 3] "); word_print(f, w); fprintf(f, "\n");}
}
+void word_buffer_print_compact(FILE *f, word_buffer *wb)
+{
+ word *w;
+ w = word_buffer_bm3(wb);
+ if(w){ fprintf(f, "%d:%s ", word_get_relative_index(w), w->form);}
+ w = word_buffer_bm2(wb);
+ if(w){ fprintf(f, "%d:%s ", word_get_relative_index(w), w->form);}
+ w = word_buffer_bm1(wb);
+ if(w){ fprintf(f, "%d:%s ", word_get_relative_index(w), w->form);}
+ w = word_buffer_b0(wb);
+ if(w){ fprintf(f, "[%d:%s] ", word_get_relative_index(w), w->form);}
+ w = word_buffer_b1(wb);
+ if(w){ fprintf(f, "%d:%s ", word_get_relative_index(w), w->form);}
+ w = word_buffer_b2(wb);
+ if(w){ fprintf(f, "%d:%s ", word_get_relative_index(w), w->form);}
+ w = word_buffer_b3(wb);
+ if(w){ fprintf(f, "%d:%s ", word_get_relative_index(w), w->form);}
+ fprintf(f, "\n");
+}
+
void word_buffer_free(word_buffer *wb)
{
int i;
@@ -121,7 +141,7 @@ int word_buffer_move_left(word_buffer *wb)
int word_buffer_end(word_buffer *wb)
{
- return (wb->current_index == wb->nbelem)? 1 : 0;
+ return (wb->current_index >= wb->nbelem)? 1 : 0;
}
int word_buffer_is_last(word_buffer *wb)
diff --git a/maca_trans_parser/src/config.c b/maca_trans_parser/src/config.c
index 43c4cdb2227a0853ac28edd9da5575cadb421f67..8e809a8b0b6b6ac1d85a232f97b6abb8a6245b42 100644
--- a/maca_trans_parser/src/config.c
+++ b/maca_trans_parser/src/config.c
@@ -106,14 +106,14 @@ void config_print(FILE *f, config *c)
word *s0 = NULL;
if(c){
if(!stack_is_empty(c->st))
- s0 = stack_elt_n(c->st, 0);
+ s0 = stack_elt_n(c->st, 0);
b0 = word_buffer_b0(c->bf);
- if(s0) { printf("s0 = "); word_print2(stdout, s0);}
- if(b0) { printf("b0 = "); word_print2(stdout, b0);}
+ /* if(s0) { printf("s0 = "); word_print2(stdout, s0);} */
+ /* if(b0) { printf("b0 = "); word_print2(stdout, b0);} */
stack_print(f, c->st);
fprintf(f, "\n");
- word_buffer_print(f, c->bf);
+ word_buffer_print_compact(f, c->bf);
}
}
diff --git a/maca_trans_parser/src/maca_trans_parser_mcf2cff.c b/maca_trans_parser/src/maca_trans_parser_mcf2cff.c
index 64cd4dfbc8bbfe9b2b365a019783d2413ca9061a..645307c5596c17747d6d2bd7574f1cdd9279a4a4 100644
--- a/maca_trans_parser/src/maca_trans_parser_mcf2cff.c
+++ b/maca_trans_parser/src/maca_trans_parser_mcf2cff.c
@@ -51,9 +51,11 @@ void generate_training_file_stream(FILE *output_file, context *ctx)
feat_vec *fv = feat_vec_new(feature_types_nb);
int sentence_nb = 0;
int root_label = dico_string2int(ctx->dico_labels, (char *) ctx->root_label);
+ int eos_label = dico_string2int(ctx->dico_labels, "eos");
word_buffer *ref = word_buffer_load_mcf(ctx->input_filename, ctx->mcd_struct);
FILE *mcf_file = myfopen(ctx->input_filename, "r");
+ /* create an mcd that corresponds to ctx->mcd_struct, but without gov and label */
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);
@@ -61,10 +63,11 @@ void generate_training_file_stream(FILE *output_file, context *ctx)
c = config_initial(mcf_file, mcd_struct_hyp, 5);
while(!word_buffer_end(ref)){
- printf("************ REF ************\n");
+ /* printf("************ REF ************\n");
word_buffer_print(stdout, ref);
- printf("*****************************\n");
+ printf("*****************************\n");*/
+ printf("*****************************\n");
config_print(stdout,c);
config2feat_vec_cff(ctx->features_model, c, ctx->d_perceptron_features, fv, ctx->mode);
@@ -74,32 +77,50 @@ void generate_training_file_stream(FILE *output_file, context *ctx)
mvt_type = movement_type(mvt_code);
mvt_label = movement_label(mvt_code);
-
- /* printf("mvt code = %d\n", mvt_code); */
+
+ /* printf("mvt code = %d\n", mvt_code);
+ printf("mvt type = %d\n", mvt_type); */
movement_print(stdout, mvt_code, ctx->dico_labels);
fprintf(output_file, "%d", mvt_code);
feat_vec_print(output_file, fv);
-
- if((mvt_type == MVT_RIGHT) && (mvt_label == root_label)){ /* sentence is complete */
-
- /* create the root arc */
- movement_right_arc(c, mvt_label, 0);
-
- /* pop root from stack */
- stack_pop(config_get_stack(c));
-
- /* printf("sentence complete config : ");
- config_print(stdout,c); */
- }
+
if(mvt_type == MVT_LEFT){
movement_left_arc(c, mvt_label, 0);
continue;
}
+
if(mvt_type == MVT_RIGHT){
movement_right_arc(c, mvt_label, 0);
word_buffer_move_right(ref);
+
+ if((mvt_label == eos_label)){ /* sentence is complete */
+ printf("sentence complete\n");
+
+ while(movement_reduce(c,0)){
+ printf("reduce\n");
+ printf("*****************************\n");
+ config_print(stdout,c);
+
+ }
+
+ /* pop eos from stack */
+ /* stack_pop(config_get_stack(c)); */
+ /* pop root from stack */
+ /* stack_pop(config_get_stack(c)); */
+
+ config_print(stdout,c);
+
+ printf("ref current index = %d\n", word_buffer_get_current_index(ref));
+ if(word_buffer_is_last(ref)){
+ printf("it is the end\n");
+ break;
+ }
+
+ /* change index of dummy word */
+ word_set_relative_index(stack_top(config_get_stack(c)), word_get_relative_index(word_buffer_b0(config_get_buffer(c))) - 1);
+ }
continue;
}
if(mvt_type == MVT_REDUCE){
diff --git a/maca_trans_parser/src/movement_parser_arc_eager.c b/maca_trans_parser/src/movement_parser_arc_eager.c
index c1f8a207bd46315ccc791aa672a7b93c62e541fc..56beba1d06b79bd3b91f19a8cbfa3125b630e9b1 100644
--- a/maca_trans_parser/src/movement_parser_arc_eager.c
+++ b/maca_trans_parser/src/movement_parser_arc_eager.c
@@ -21,7 +21,7 @@ int movement_type(int mvt)
if(mvt == 0) return MVT_SHIFT; /* 0 is the code of shift */
if(mvt == 1) return MVT_REDUCE; /* 1 is the code of reduce */
if(mvt % 2 == 0) return MVT_LEFT; /* even movements are left movements */
- return MVT_LEFT; /* odd movements are right movements */
+ return MVT_RIGHT; /* odd movements are right movements */
}
int movement_label(int mvt)
@@ -40,15 +40,21 @@ int movement_left_arc(config *c, int label, float score)
if(word_buffer_is_empty(c->bf)) return 0;
/* word on top of the stack should not have a governor */
-
- printf("word_get_gov_relative_index(stack_top(c->st)) = %d\n", word_get_gov_relative_index(stack_top(c->st)));
+ /* printf("index word top of stack = %d\n", word_get_relative_index(stack_top(c->st))); */
+ /* printf("word_get_gov_relative_index(stack_top(c->st)) = %d\n", word_get_gov_relative_index(stack_top(c->st))); */
if(word_get_gov_relative_index(stack_top(c->st)) != -1) return 0;
+ word *gov = word_buffer_b0(c->bf);
+ word *dep = stack_top(c->st);
+ int dist = (word_get_relative_index(gov)) - (word_get_relative_index(dep));
+
+ printf("create left arc %d <- %d dist = %d\n", word_get_relative_index(dep), word_get_relative_index(gov), dist);
+
/* create a new dependency */
- word_set_gov(stack_top(c->st), word_get_relative_index(word_buffer_b0(c->bf)));
- word_set_label(stack_top(c->st), label);
+ word_set_gov(dep, dist);
+ word_set_label(dep, label);
/* depset_add(c->ds, word_buffer_b0(c->bf), label, stack_top(c->st)); */
stack_pop(c->st);
@@ -62,12 +68,17 @@ int movement_right_arc(config *c, int label, float score)
if(stack_is_empty(c->st)) return 0;
if(word_buffer_is_empty(c->bf)) return 0;
+ word *gov = stack_top(c->st);
+ word *dep = word_buffer_b0(c->bf);
+ int dist = (word_get_relative_index(gov)) - (word_get_relative_index(dep));
+
+ printf("create right arc %d -> %d dist = %d\n", word_get_relative_index(gov), word_get_relative_index(dep), dist);
+
/* create a new dependency */
- word_set_gov(word_buffer_b0(c->bf), word_get_relative_index(stack_top(c->st)));
- word_set_label(word_buffer_b0(c->bf), label);
+ word_set_gov(dep, dist);
+ word_set_label(dep, label);
- /* depset_add(c->ds, stack_top(c->st), label, word_buffer_b0(c->bf)); */
stack_push(c->st, word_buffer_b0(c->bf));
word_buffer_move_right(c->bf);
@@ -91,7 +102,7 @@ int movement_shift(config *c, int stream, float score)
int movement_reduce(config *c, float score)
{
if(stack_is_empty(c->st)) return 0;
- if(word_get_gov(stack_top(c->st)) == -1) return 0; /* word on top of stack does not have a governor */
+ if(word_get_gov(stack_top(c->st)) == 0) return 0; /* word on top of stack does not have a governor */
stack_pop(c->st);
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 f719363d27f95d661d1ed3efb2e26a18878e2e6e..a24442bf116d7df7ed426a1f45ae4a091ae565d6 100644
--- a/maca_trans_parser/src/oracle_parser_arc_eager.c
+++ b/maca_trans_parser/src/oracle_parser_arc_eager.c
@@ -7,12 +7,18 @@
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;
for(dep=1; dep < ref->nbelem; dep++){
- if(word_get_gov_relative_index(word_buffer_get_word_n(ref, dep)) == word_index){ /* found a dependent of word in ref */
+ gov_ref = word_get_gov_relative_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 */
-
- if(word_get_gov_relative_index(word_buffer_get_word_n(config_get_buffer(c), dep)) != word_index) return 0;
+ printf("found a dep of word %d in ref, it is %d\n", word_index, dep);
+ gov_hyp = word_get_gov_relative_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)
@@ -29,38 +35,42 @@ int oracle_parser_arc_eager(config *c, word_buffer *ref)
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;
if(!stack_is_empty(c->st) && !word_buffer_is_empty(c->bf)){
s0 = stack_top(c->st);
s0_index = word_get_relative_index(s0);
+ s0_gov_index = word_get_gov_relative_index(word_buffer_get_word_n(ref, s0_index));
b0 = word_buffer_b0(c->bf);
b0_index = word_get_relative_index(b0);
+ b0_gov_index = word_get_gov_relative_index(word_buffer_get_word_n(ref, b0_index));
- printf("s0_index = %d b0_index = %d\n", s0_index, b0_index);
- printf("dans ref gov de s0 (%d) = %d\n", s0_index, word_get_gov_relative_index(word_buffer_get_word_n(ref, s0_index)));
- printf("dans ref gov de b0 (%d) = %d\n", b0_index, word_get_gov_relative_index(word_buffer_get_word_n(ref, b0_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);*/
/* LEFT ARC b0 is the governor and s0 the dependent */
- if(word_get_gov_relative_index(word_buffer_get_word_n(ref, s0_index)) == b0_index){
+ if(s0_gov_index == b0_index){
- printf("oracle says left\n");
+ /* printf("oracle says left\n"); */
return movement_left_code(word_get_label(word_buffer_get_word_n(ref, s0_index)));
}
/* RIGHT ARC s0 is the governor and b0 the dependent */
- if((word_get_gov_relative_index(word_buffer_get_word_n(ref, b0_index)) == s0_index))
+ if(b0_gov_index == s0_index){
+ /* printf("oracle says right\n"); */
return movement_right_code(word_get_label(word_buffer_get_word_n(ref, b0_index)));
-
+ }
/* REDUCE */
- if((stack_height(c->st) > 2) && check_all_dependents_of_word_in_ref_are_in_hyp(c, ref, s0_index))
-
- /* if(word_get_gov_relative_index(stack_top(c->st)) != -1) */
- return MVT_REDUCE;
-
- /* if(check_all_dependents_of_word_in_ref_are_in_hyp(c, ref, s0_index))
- return MVT_REDUCE;*/
+ printf("all dep in ref are in hyp = %d\n", check_all_dependents_of_word_in_ref_are_in_hyp(c, ref, s0_index));
+ if((stack_height(c->st) > 2)
+ && check_all_dependents_of_word_in_ref_are_in_hyp(c, ref, s0_index)
+ && (word_get_gov(stack_top(c->st)) != 0)) /* word on top of the stack has a goveror */
+ {
+ return MVT_REDUCE;
+ }
/* SHIFT */
return MVT_SHIFT;