Add PH_Stats().

src/ph.c

@@ -437,6 +437,38 @@ PH_Dump(struct ph *ph, char **strings)
 	return (sb);
 }
 
+void
+PH_Stats(const struct ph * const restrict ph,
+	 char * const restrict * const restrict strings,
+	 struct ph_stats * const restrict stats)
+{
+	CHECK_OBJ_NOTNULL(stats, PH_STATS_MAGIC);
+
+	memset(stats, 0, sizeof(*stats));
+	if (ph == NULL)
+		return;
+	CHECK_OBJ(ph, PH_MAGIC);
+	CHECK_OBJ_NOTNULL(ph->h1, HASH_MAGIC);
+	AN(ph->tbl);
+	AN(ph->collision);
+	AN(strings);
+
+	stats->buckets = ph->h1->mask + 1;
+	stats->klen = ph->l;
+	stats->minlen = ph->minlen;
+	stats->maxlen = ph->maxlen;
+
+	for (unsigned i = 0; i <= ph->h1->mask; i++)
+		if (vbit_test(ph->collision, i)) {
+			struct hash *h2;
+
+			h2 = ph->tbl[i].h2;
+			CHECK_OBJ_NOTNULL(h2, HASH_MAGIC);
+
+			stats->collisions++;
+		}
+}
+
 void
 PH_Free(struct ph *ph)
 {

src/ph.h

@@ -42,6 +42,16 @@
  */
 struct ph;
 
+struct ph_stats {
+	unsigned	magic;
+#define PH_STATS_MAGIC 0x68b803bb
+	uint64_t	buckets;
+	uint64_t	collisions;
+	uint64_t	klen;
+	uint64_t	minlen;
+	uint64_t	maxlen;
+};
+
 /*
  * Initialize perfect hashing. Supplies a seed for random number
  * generation, which should be obtained from an entropy source.  Only
@@ -76,6 +86,10 @@ unsigned PH_Lookup(const struct ph * const restrict ph,
 		   char * const restrict * const restrict strings,
 		   const char * const restrict subject);
 
+void PH_Stats(const struct ph * const restrict ph,
+	      char * const restrict * const restrict strings,
+	      struct ph_stats * const restrict stats);
+
 /*
  * Return a string dump of ph as generated for strings.
  *

src/tests/bench/bench_ph.c

@@ -115,6 +115,7 @@ main(int argc, char *argv[])
 	struct timespec before, after, start, finish;
 	uint64_t ns = 0, iters, matches;
 	uint32_t seed[4];
+	struct ph_stats stats = { .magic = PH_STATS_MAGIC };
 	int opt, do_shuf = 0, do_iters = ITERATIONS;
 	struct rusage rusage;
 
@@ -314,6 +315,17 @@ main(int argc, char *argv[])
 		printf("... done.\n");
 	}
 
+	printf("\nGetting stats ...\n");
+	(void)clock_gettime(CLOCK, &before);
+	PH_Stats(ph, strings, &stats);
+	(void)clock_gettime(CLOCK, &after);
+	printf("Stats computed in %lu ns\n", tdiff(&before, &after));
+	printf("%lu buckets\n", stats.buckets);
+	printf("%lu collisions\n", stats.collisions);
+	printf("%lu key vector length\n", stats.klen);
+	printf("%lu min string length\n", stats.minlen);
+	printf("%lu max string length\n", stats.maxlen);
+
 	if (do_iters == 0)
 		exit(EXIT_SUCCESS);