bugfix for the building of bottom-up filter in the C runtime

2012-11-16 13:27:15 +00:00
parent 7d616c01b4
commit 954d7a7ff5
1 changed files with 141 additions and 75 deletions
--- a/src/runtime/c/pgf/parser.c
+++ b/src/runtime/c/pgf/parser.c
@@ -1971,6 +1971,8 @@ pgf_parser_leftcorner_add_token(PgfConcr* concr,
 	}

 	PgfCFCat cfc = {item->conts->ccat, item->conts->lin_idx};
+	while (cfc.ccat->conts != NULL)
+		cfc.ccat = cfc.ccat->conts->ccat;
 	PgfProductionSeq prods = gu_map_get(set, &cfc, PgfProductionSeq);

 	if (gu_seq_length(item->args) == 0) {
@@ -1993,6 +1995,8 @@ pgf_parser_leftcorner_add_epsilon(PgfConcr* concr,
                                  GuPool *pool)
 {
 	PgfCFCat cfc = {item->conts->ccat, item->conts->lin_idx};
+	while (cfc.ccat->conts != NULL)
+		cfc.ccat = cfc.ccat->conts->ccat;
 	PgfProductionSeq prods =
 		gu_map_get(concr->epsilon_idx, &cfc, PgfProductionSeq);

@@ -2041,16 +2045,32 @@ pgf_parser_leftcorner_item(PgfLeftcornerFn* clo, PgfItem* item)
 			case PGF_SYMBOL_CAT: {
 				PgfSymbolCat* scat = i.data;

-				// Place the item in the continuation map
 				PgfPArg* parg =
 					gu_seq_index(item->args, PgfPArg, scat->d);

+				if (gu_seq_is_null(parg->ccat->prods))
+					return;
+
+				// Place the item in the continuation map
 				PgfItemConts* conts_ =
 					pgf_parsing_get_conts(clo->conts_map,
 										  parg->ccat, scat->r, 0,
 										  clo->tmp_pool);
 				gu_buf_push(conts_->items, PgfItem*, item);
+				if (gu_buf_length(conts_->items) == 1) {
+					conts_->outside_prob = 0;

+					size_t n_prods = gu_seq_length(parg->ccat->prods);
+					for (size_t i = 0; i < n_prods; i++) {
+						PgfProduction prod =
+							gu_seq_get(parg->ccat->prods, PgfProduction, i);
+
+						PgfItem* item =
+							pgf_new_item(conts_, prod, clo->tmp_pool, NULL);
+
+						pgf_parser_leftcorner_item(clo, item);
+					}
+				} else {
 					// If the current category has epsilon rules
 					// then we must do the same for a new item where 
 					// the dot is moved with one position.
@@ -2063,6 +2083,7 @@ pgf_parser_leftcorner_item(PgfLeftcornerFn* clo, PgfItem* item)
 						pgf_item_advance(new_item, clo->tmp_pool);
 						pgf_parser_leftcorner_item(clo, new_item);
 					}
+				}
 				break;
 			}
 			case PGF_SYMBOL_KS: {
@@ -2101,10 +2122,11 @@ pgf_parser_leftcorner_item(PgfLeftcornerFn* clo, PgfItem* item)
 			if (eps_ccat == NULL) {
 				eps_ccat = gu_new(PgfCCat, clo->tmp_pool);
 				eps_ccat->cnccat = item->conts->ccat->cnccat;
-				eps_ccat->viterbi_prob = INFINITY;
+				eps_ccat->viterbi_prob = 0;
 				eps_ccat->fid = clo->max_fid++;
 				eps_ccat->prods =
 					gu_buf_seq(gu_new_buf(PgfProduction, clo->tmp_pool));
+				eps_ccat->conts = item->conts;
 				eps_ccat->n_synprods = 0;

 				gu_map_put(clo->generated_cats, 
@@ -2139,6 +2161,8 @@ pgf_parser_leftcorner_item(PgfLeftcornerFn* clo, PgfItem* item)
 				for (size_t i = 0; i < n_items; i++) {
 					PgfItem* cont = 
 						gu_buf_get(item->conts->items, PgfItem*, i);
+					if (cont == NULL)
+						continue;
 					
 					gu_assert(gu_variant_tag(cont->curr_sym) == PGF_SYMBOL_CAT);
 					PgfSymbolCat* scat = gu_variant_data(cont->curr_sym);
@@ -2153,21 +2177,45 @@ pgf_parser_leftcorner_item(PgfLeftcornerFn* clo, PgfItem* item)
 	}
 	break;
 	case PGF_PRODUCTION_COERCE: {
+		if (item->seq_idx < 1) {
 			PgfPArg* parg =
 				gu_seq_index(item->args, PgfPArg, 0);

+			if (gu_seq_is_null(parg->ccat->prods))
+				return;
+
 			PgfItemConts* conts_ =
 				pgf_parsing_get_conts(clo->conts_map,
 									  parg->ccat, item->conts->lin_idx, 0,
 									  clo->tmp_pool);
 			gu_buf_push(conts_->items, PgfItem*, item);
+			if (gu_buf_length(conts_->items) == 1) {
+				conts_->outside_prob = 0;

+				size_t n_prods = gu_seq_length(parg->ccat->prods);
+				for (size_t i = 0; i < n_prods; i++) {
+					PgfProduction prod =
+						gu_seq_get(parg->ccat->prods, PgfProduction, i);
+
+					PgfItem* new_item =
+						pgf_new_item(conts_, prod, clo->tmp_pool, NULL);
+
+					pgf_parser_leftcorner_item(clo, new_item);
+				}
+			} else {
 				// If the argument category has epsilon rules
 				// then the result category has epsilon rules too.
-		PgfCCat* eps_arg_ccat =
+				PgfCCat* eps_ccat =
 					gu_map_get(clo->generated_cats, conts_, PgfCCat*);

-		if (eps_arg_ccat != NULL) {
+				if (eps_ccat != NULL) {
+					PgfItem* new_item =
+						pgf_item_update_arg(item, 0, eps_ccat, clo->tmp_pool, NULL);
+					pgf_item_advance(new_item, clo->tmp_pool);
+					pgf_parser_leftcorner_item(clo, new_item);
+				}
+			}
+		} else {
 			PgfCCat* tmp_ccat =
 				gu_map_get(clo->generated_cats, item->conts, PgfCCat*);

@@ -2175,21 +2223,23 @@ pgf_parser_leftcorner_item(PgfLeftcornerFn* clo, PgfItem* item)
 			if (eps_res_ccat == NULL) {
 				eps_res_ccat = gu_new(PgfCCat, clo->tmp_pool);
 				eps_res_ccat->cnccat = item->conts->ccat->cnccat;
-				eps_res_ccat->viterbi_prob = INFINITY;
+				eps_res_ccat->viterbi_prob = 0;
 				eps_res_ccat->fid = clo->max_fid++;
 				eps_res_ccat->prods =
 					gu_buf_seq(gu_new_buf(PgfProduction, clo->tmp_pool));
+				eps_res_ccat->conts = item->conts;
 				eps_res_ccat->n_synprods = 0;

 				gu_map_put(clo->generated_cats, 
 						   item->conts, PgfCCat*, eps_res_ccat);
+			}

 			PgfProduction prod;
 			PgfProductionCoerce* new_pcoerce =
 				gu_new_variant(PGF_PRODUCTION_COERCE,
 							   PgfProductionCoerce,
 							   &prod, clo->pool);
-				new_pcoerce->coerce = eps_arg_ccat;
+			new_pcoerce->coerce = gu_seq_index(item->args, PgfPArg, 0)->ccat;
 			GuBuf* prodbuf = gu_seq_buf(eps_res_ccat->prods);
 			gu_buf_push(prodbuf, PgfProduction, prod);

@@ -2216,6 +2266,8 @@ pgf_parser_leftcorner_item(PgfLeftcornerFn* clo, PgfItem* item)
 				for (size_t i = 0; i < n_items; i++) {
 					PgfItem* cont = 
 						gu_buf_get(item->conts->items, PgfItem*, i);
+					if (cont == NULL)
+						continue;

 					gu_assert(gu_variant_tag(cont->curr_sym) == PGF_SYMBOL_CAT);
 					PgfSymbolCat* scat = gu_variant_data(cont->curr_sym);
@@ -2227,7 +2279,6 @@ pgf_parser_leftcorner_item(PgfLeftcornerFn* clo, PgfItem* item)
 				}
 			}
 		}
-		}
 		break;
 	}
 	default:
@@ -2246,19 +2297,25 @@ pgf_parser_leftcorner_iter_cats(GuMapItor* fn, const void* key, void* value, GuE
 	if (gu_seq_is_null(ccat->prods))
 		return;

+	for (size_t lin_idx = 0; lin_idx < ccat->cnccat->n_lins; lin_idx++) {
+		PgfItemConts* conts =
+			pgf_parsing_get_conts(clo->conts_map, ccat, lin_idx, 0, clo->tmp_pool);
+				
+		gu_buf_push(conts->items, PgfItem*, NULL);
+		if (gu_buf_length(conts->items) == 1) {
+			conts->outside_prob = 0;
+
 			size_t n_prods = gu_seq_length(ccat->prods);
 			for (size_t i = 0; i < n_prods; i++) {
 				PgfProduction prod = gu_seq_get(ccat->prods, PgfProduction, i);

-		for (size_t lin_idx = 0; lin_idx < ccat->cnccat->n_lins; lin_idx++) {
-			PgfItemConts* conts =
-				pgf_parsing_get_conts(clo->conts_map, ccat, lin_idx, 0, clo->tmp_pool);
 				PgfItem* item =
 					pgf_new_item(conts, prod, clo->tmp_pool, NULL);

 				pgf_parser_leftcorner_item(clo, item);
 			}
 		}
+	}
 }

 static void
@@ -2268,8 +2325,13 @@ pgf_parser_leftcorner_closure(PgfProductionIdx* set, PgfItemBuf* items,
 	size_t n_items = gu_buf_length(items);
 	for (size_t i = 0; i < n_items; i++) {
 		PgfItem* item = gu_buf_get(items, PgfItem*, i);
+		if (item == NULL)
+			continue;

 		PgfCFCat cfc = {item->conts->ccat, item->conts->lin_idx};
+		while (cfc.ccat->conts != NULL)
+			cfc.ccat = cfc.ccat->conts->ccat;
+
 		if (!gu_map_has(set, &cfc)) {
 			gu_map_put(set, &cfc, PgfCCat*, NULL);

@@ -2294,12 +2356,16 @@ pgf_parser_leftcorner_iter_conts(GuMapItor* fn, const void* key, void* value, Gu
 	PgfCCat* ccat = (PgfCCat*) key;
 	PgfItemContss* contss = *((PgfItemContss**) value);

+	if (ccat->conts != NULL)
+		return;
+
 	size_t n_lins = gu_list_length(contss);
 	for (size_t lin_idx = 0; lin_idx < n_lins; lin_idx++) {
 		PgfItemConts* conts = gu_list_index(contss, lin_idx);

 		if (conts != NULL) {
 			PgfCFCat cfc = {ccat, lin_idx};
+
 			PgfProductionIdx* set = 
 				gu_map_get(concr->leftcorner_cat_idx, 
 		                   &cfc, PgfProductionIdx*);