Merge pull request #1518 from ampli/disjunct-count

eliminate_duplicate_disjuncts(): Return the discarded disjunct count
opencog · May 3, 2024 · ea8c04e · ea8c04e
2 parents f393ef5 + a9f081c
commit ea8c04e
Show file tree

Hide file tree

Showing 6 changed files with 23 additions and 13 deletions.
diff --git a/link-grammar/dict-common/print-dict.c b/link-grammar/dict-common/print-dict.c
@@ -822,8 +822,7 @@ static char *display_disjuncts(Dictionary dict, const Dict_node *dn,
 		                                      max_cost, NULL);
 
 		unsigned int dnum0 = count_disjuncts(d);
-		d = eliminate_duplicate_disjuncts(d, false);
-		unsigned int dnum1 = count_disjuncts(d);
+		unsigned int dnum1 = dnum0 - eliminate_duplicate_disjuncts(d, false);
 
 		if ((flags != NULL) && (strchr(flags, 'm') != NULL))
 		{

diff --git a/link-grammar/disjunct-utils.c b/link-grammar/disjunct-utils.c
@@ -334,10 +334,15 @@ static void disjunct_dup_table_delete(disjunct_dup_table *dt)
 }
 
 /**
- * Takes the list of disjuncts pointed to by d, eliminates all
- * duplicates, and returns a pointer to a new list.
+ * Takes the list of disjuncts pointed to by dw, eliminates all
+ * duplicates. The elimination is done in-place. Because the first
+ * disjunct can never be eliminated (it cannot be a duplicate of
+ * anything before it), the argument disjunct list always points to the
+ * new list.
+ *
+ * @return The number of eliminated disjuncts.
  */
-Disjunct *eliminate_duplicate_disjuncts(Disjunct *dw, bool multi_string)
+unsigned int eliminate_duplicate_disjuncts(Disjunct *dw, bool multi_string)
 {
 	unsigned int count = 0;
 	disjunct_dup_table *dt;
@@ -408,7 +413,7 @@ Disjunct *eliminate_duplicate_disjuncts(Disjunct *dw, bool multi_string)
 	        multi_string ? " (different word-strings)" : "");
 
 	disjunct_dup_table_delete(dt);
-	return dw;
+	return count;
 }
 
 /* ============================================================= */

diff --git a/link-grammar/disjunct-utils.h b/link-grammar/disjunct-utils.h
@@ -85,7 +85,7 @@ void free_categories(Sentence);
 void free_categories_from_disjunct_array(Disjunct *, unsigned int);
 unsigned int count_disjuncts(Disjunct *);
 Disjunct * catenate_disjuncts(Disjunct *, Disjunct *);
-Disjunct * eliminate_duplicate_disjuncts(Disjunct *, bool);
+unsigned int eliminate_duplicate_disjuncts(Disjunct *, bool);
 int left_connector_count(Disjunct *);
 int right_connector_count(Disjunct *);
 

diff --git a/link-grammar/parse/preparation.c b/link-grammar/parse/preparation.c
@@ -166,8 +166,10 @@ static void create_wildcard_word_disjunct_list(Sentence sent,
 	build_sentence_disjuncts(wc_word_list, opts->disjunct_cost, opts);
 
 	Word *word0 = &wc_word_list->word[0];
-	word0->d = eliminate_duplicate_disjuncts(word0->d, false);
-	word0->d = eliminate_duplicate_disjuncts(word0->d, true);
+	unsigned int Ndeleted;
+	Ndeleted = eliminate_duplicate_disjuncts(word0->d, false);
+	Ndeleted += eliminate_duplicate_disjuncts(word0->d, true);
+	print_time(opts, "Eliminated duplicate disjuncts (%u deleted)", Ndeleted);
 
 	wc_word_list->min_len_encoding = 2; /* Don't share/encode. */
 	Tracon_sharing *t = pack_sentence_for_pruning(wc_word_list);
@@ -208,15 +210,16 @@ void prepare_to_parse(Sentence sent, Parse_Options opts)
 	}
 	print_time(opts, "Built disjuncts");
 
+	unsigned int Ndeleted = 0;
 	for (i=0; i<sent->length; i++)
 	{
-		sent->word[i].d = eliminate_duplicate_disjuncts(sent->word[i].d, false);
+		Ndeleted += eliminate_duplicate_disjuncts(sent->word[i].d, false);
 		if (IS_GENERATION(sent->dict))
 		{
 			if ((sent->word[i].d != NULL) && (sent->word[i].d->is_category != 0))
 			{
 				/* Also with different word_string. */
-				sent->word[i].d = eliminate_duplicate_disjuncts(sent->word[i].d, true);
+				Ndeleted += eliminate_duplicate_disjuncts(sent->word[i].d, true);
 
 				/* XXX This ordinal numbering is just plain wrong.
 				 * Most of the disjuncts have already been pruned away,
@@ -248,7 +251,7 @@ void prepare_to_parse(Sentence sent, Parse_Options opts)
 			return;
 #endif
 	}
-	print_time(opts, "Eliminated duplicate disjuncts");
+	print_time(opts, "Eliminated duplicate disjuncts (%u deleted)", Ndeleted);
 
 	if (verbosity_level(D_PREP))
 	{

diff --git a/link-grammar/resources.c b/link-grammar/resources.c
@@ -145,7 +145,7 @@ bool resources_memory_exhausted(Resources r)
 	else return (r->memory_exhausted || (get_space_in_use() > r->max_memory));
 }
 
-#define RES_COL_WIDTH 40
+#define RES_COL_WIDTH 52
 
 /** print out the cpu ticks since this was last called */
 GNUC_PRINTF(2,0)

diff --git a/link-grammar/tokenize/word-structures.h b/link-grammar/tokenize/word-structures.h
@@ -14,6 +14,9 @@
 #ifndef _WORD_STRUCTURE_H_
 #define _WORD_STRUCTURE_H_
 
+#include <inttypes.h>
+#include "stdbool.h"
+
 #include "api-types.h"
 
 typedef struct X_node_struct X_node;