summaryrefslogtreecommitdiff
path: root/lib/libc/stdlib/heapsort.c
diff options
context:
space:
mode:
authorTheo de Raadt <deraadt@cvs.openbsd.org>1995-10-18 08:53:40 +0000
committerTheo de Raadt <deraadt@cvs.openbsd.org>1995-10-18 08:53:40 +0000
commitd6583bb2a13f329cf0332ef2570eb8bb8fc0e39c (patch)
treeece253b876159b39c620e62b6c9b1174642e070e /lib/libc/stdlib/heapsort.c
initial import of NetBSD tree
Diffstat (limited to 'lib/libc/stdlib/heapsort.c')
-rw-r--r--lib/libc/stdlib/heapsort.c184
1 files changed, 184 insertions, 0 deletions
diff --git a/lib/libc/stdlib/heapsort.c b/lib/libc/stdlib/heapsort.c
new file mode 100644
index 00000000000..3be56ba1d64
--- /dev/null
+++ b/lib/libc/stdlib/heapsort.c
@@ -0,0 +1,184 @@
+/*-
+ * Copyright (c) 1991, 1993
+ * The Regents of the University of California. All rights reserved.
+ *
+ * This code is derived from software contributed to Berkeley by
+ * Ronnie Kon at Mindcraft Inc., Kevin Lew and Elmer Yglesias.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ * 3. All advertising materials mentioning features or use of this software
+ * must display the following acknowledgement:
+ * This product includes software developed by the University of
+ * California, Berkeley and its contributors.
+ * 4. Neither the name of the University nor the names of its contributors
+ * may be used to endorse or promote products derived from this software
+ * without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ */
+
+#if defined(LIBC_SCCS) && !defined(lint)
+/*static char sccsid[] = "from: @(#)heapsort.c 8.1 (Berkeley) 6/4/93";*/
+static char *rcsid = "$Id: heapsort.c,v 1.1 1995/10/18 08:42:17 deraadt Exp $";
+#endif /* LIBC_SCCS and not lint */
+
+#include <sys/types.h>
+#include <errno.h>
+#include <stdlib.h>
+
+/*
+ * Swap two areas of size number of bytes. Although qsort(3) permits random
+ * blocks of memory to be sorted, sorting pointers is almost certainly the
+ * common case (and, were it not, could easily be made so). Regardless, it
+ * isn't worth optimizing; the SWAP's get sped up by the cache, and pointer
+ * arithmetic gets lost in the time required for comparison function calls.
+ */
+#define SWAP(a, b, count, size, tmp) { \
+ count = size; \
+ do { \
+ tmp = *a; \
+ *a++ = *b; \
+ *b++ = tmp; \
+ } while (--count); \
+}
+
+/* Copy one block of size size to another. */
+#define COPY(a, b, count, size, tmp1, tmp2) { \
+ count = size; \
+ tmp1 = a; \
+ tmp2 = b; \
+ do { \
+ *tmp1++ = *tmp2++; \
+ } while (--count); \
+}
+
+/*
+ * Build the list into a heap, where a heap is defined such that for
+ * the records K1 ... KN, Kj/2 >= Kj for 1 <= j/2 <= j <= N.
+ *
+ * There two cases. If j == nmemb, select largest of Ki and Kj. If
+ * j < nmemb, select largest of Ki, Kj and Kj+1.
+ */
+#define CREATE(initval, nmemb, par_i, child_i, par, child, size, count, tmp) { \
+ for (par_i = initval; (child_i = par_i * 2) <= nmemb; \
+ par_i = child_i) { \
+ child = base + child_i * size; \
+ if (child_i < nmemb && compar(child, child + size) < 0) { \
+ child += size; \
+ ++child_i; \
+ } \
+ par = base + par_i * size; \
+ if (compar(child, par) <= 0) \
+ break; \
+ SWAP(par, child, count, size, tmp); \
+ } \
+}
+
+/*
+ * Select the top of the heap and 'heapify'. Since by far the most expensive
+ * action is the call to the compar function, a considerable optimization
+ * in the average case can be achieved due to the fact that k, the displaced
+ * elememt, is ususally quite small, so it would be preferable to first
+ * heapify, always maintaining the invariant that the larger child is copied
+ * over its parent's record.
+ *
+ * Then, starting from the *bottom* of the heap, finding k's correct place,
+ * again maintianing the invariant. As a result of the invariant no element
+ * is 'lost' when k is assigned its correct place in the heap.
+ *
+ * The time savings from this optimization are on the order of 15-20% for the
+ * average case. See Knuth, Vol. 3, page 158, problem 18.
+ *
+ * XXX Don't break the #define SELECT line, below. Reiser cpp gets upset.
+ */
+#define SELECT(par_i, child_i, nmemb, par, child, size, k, count, tmp1, tmp2) { \
+ for (par_i = 1; (child_i = par_i * 2) <= nmemb; par_i = child_i) { \
+ child = base + child_i * size; \
+ if (child_i < nmemb && compar(child, child + size) < 0) { \
+ child += size; \
+ ++child_i; \
+ } \
+ par = base + par_i * size; \
+ COPY(par, child, count, size, tmp1, tmp2); \
+ } \
+ for (;;) { \
+ child_i = par_i; \
+ par_i = child_i / 2; \
+ child = base + child_i * size; \
+ par = base + par_i * size; \
+ if (child_i == 1 || compar(k, par) < 0) { \
+ COPY(child, k, count, size, tmp1, tmp2); \
+ break; \
+ } \
+ COPY(child, par, count, size, tmp1, tmp2); \
+ } \
+}
+
+/*
+ * Heapsort -- Knuth, Vol. 3, page 145. Runs in O (N lg N), both average
+ * and worst. While heapsort is faster than the worst case of quicksort,
+ * the BSD quicksort does median selection so that the chance of finding
+ * a data set that will trigger the worst case is nonexistent. Heapsort's
+ * only advantage over quicksort is that it requires little additional memory.
+ */
+int
+heapsort(vbase, nmemb, size, compar)
+ void *vbase;
+ size_t nmemb, size;
+ int (*compar) __P((const void *, const void *));
+{
+ register int cnt, i, j, l;
+ register char tmp, *tmp1, *tmp2;
+ char *base, *k, *p, *t;
+
+ if (nmemb <= 1)
+ return (0);
+
+ if (!size) {
+ errno = EINVAL;
+ return (-1);
+ }
+
+ if ((k = malloc(size)) == NULL)
+ return (-1);
+
+ /*
+ * Items are numbered from 1 to nmemb, so offset from size bytes
+ * below the starting address.
+ */
+ base = (char *)vbase - size;
+
+ for (l = nmemb / 2 + 1; --l;)
+ CREATE(l, nmemb, i, j, t, p, size, cnt, tmp);
+
+ /*
+ * For each element of the heap, save the largest element into its
+ * final slot, save the displaced element (k), then recreate the
+ * heap.
+ */
+ while (nmemb > 1) {
+ COPY(k, base + nmemb * size, cnt, size, tmp1, tmp2);
+ COPY(base + nmemb * size, base + size, cnt, size, tmp1, tmp2);
+ --nmemb;
+ SELECT(i, j, nmemb, t, p, size, k, cnt, tmp1, tmp2);
+ }
+ free(k);
+ return (0);
+}