radixsort.c revision 1.2 1 /*-
2 * Copyright (c) 1990 The Regents of the University of California.
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgement:
15 * This product includes software developed by the University of
16 * California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 */
33
34 #if defined(LIBC_SCCS) && !defined(lint)
35 /*static char sccsid[] = "from: @(#)radixsort.c 5.7 (Berkeley) 2/23/91";*/
36 static char rcsid[] = "$Id: radixsort.c,v 1.2 1993/08/01 18:36:59 mycroft Exp $";
37 #endif /* LIBC_SCCS and not lint */
38
39 #include <sys/types.h>
40 #include <limits.h>
41 #include <stdlib.h>
42 #include <stddef.h>
43 #include <string.h>
44
45 /*
46 * __rspartition is the cutoff point for a further partitioning instead
47 * of a shellsort. If it changes check __rsshell_increments. Both of
48 * these are exported, as the best values are data dependent.
49 */
50 #define NPARTITION 40
51 int __rspartition = NPARTITION;
52 int __rsshell_increments[] = { 4, 1, 0, 0, 0, 0, 0, 0 };
53
54 /*
55 * Stackp points to context structures, where each structure schedules a
56 * partitioning. Radixsort exits when the stack is empty.
57 *
58 * If the buckets are placed on the stack randomly, the worst case is when
59 * all the buckets but one contain (npartitions + 1) elements and the bucket
60 * pushed on the stack last contains the rest of the elements. In this case,
61 * stack growth is bounded by:
62 *
63 * limit = (nelements / (npartitions + 1)) - 1;
64 *
65 * This is a very large number, 52,377,648 for the maximum 32-bit signed int.
66 *
67 * By forcing the largest bucket to be pushed on the stack first, the worst
68 * case is when all but two buckets each contain (npartitions + 1) elements,
69 * with the remaining elements split equally between the first and last
70 * buckets pushed on the stack. In this case, stack growth is bounded when:
71 *
72 * for (partition_cnt = 0; nelements > npartitions; ++partition_cnt)
73 * nelements =
74 * (nelements - (npartitions + 1) * (nbuckets - 2)) / 2;
75 * The bound is:
76 *
77 * limit = partition_cnt * (nbuckets - 1);
78 *
79 * This is a much smaller number, 4590 for the maximum 32-bit signed int.
80 */
81 #define NBUCKETS (UCHAR_MAX + 1)
82
83 typedef struct _stack {
84 const u_char **bot;
85 int indx, nmemb;
86 } CONTEXT;
87
88 #define STACKPUSH { \
89 stackp->bot = p; \
90 stackp->nmemb = nmemb; \
91 stackp->indx = indx; \
92 ++stackp; \
93 }
94 #define STACKPOP { \
95 if (stackp == stack) \
96 break; \
97 --stackp; \
98 bot = stackp->bot; \
99 nmemb = stackp->nmemb; \
100 indx = stackp->indx; \
101 }
102
103 /*
104 * A variant of MSD radix sorting; see Knuth Vol. 3, page 177, and 5.2.5,
105 * Ex. 10 and 12. Also, "Three Partition Refinement Algorithms, Paige
106 * and Tarjan, SIAM J. Comput. Vol. 16, No. 6, December 1987.
107 *
108 * This uses a simple sort as soon as a bucket crosses a cutoff point,
109 * rather than sorting the entire list after partitioning is finished.
110 * This should be an advantage.
111 *
112 * This is pure MSD instead of LSD of some number of MSD, switching to
113 * the simple sort as soon as possible. Takes linear time relative to
114 * the number of bytes in the strings.
115 */
116 int
117 #if __STDC__
118 radixsort(const u_char **l1, int nmemb, const u_char *tab, u_char endbyte)
119 #else
120 radixsort(l1, nmemb, tab, endbyte)
121 const u_char **l1;
122 register int nmemb;
123 const u_char *tab;
124 u_char endbyte;
125 #endif
126 {
127 register int i, indx, t1, t2;
128 register const u_char **l2;
129 register const u_char **p;
130 register const u_char **bot;
131 register const u_char *tr;
132 CONTEXT *stack, *stackp;
133 int c[NBUCKETS + 1], max;
134 u_char ltab[NBUCKETS];
135 static void shellsort();
136
137 if (nmemb <= 1)
138 return(0);
139
140 /*
141 * T1 is the constant part of the equation, the number of elements
142 * represented on the stack between the top and bottom entries.
143 * It doesn't get rounded as the divide by 2 rounds down (correct
144 * for a value being subtracted). T2, the nelem value, has to be
145 * rounded up before each divide because we want an upper bound;
146 * this could overflow if nmemb is the maximum int.
147 */
148 t1 = ((__rspartition + 1) * (NBUCKETS - 2)) >> 1;
149 for (i = 0, t2 = nmemb; t2 > __rspartition; i += NBUCKETS - 1)
150 t2 = ((t2 + 1) >> 1) - t1;
151 if (i) {
152 if (!(stack = stackp = (CONTEXT *)malloc(i * sizeof(CONTEXT))))
153 return(-1);
154 } else
155 stack = stackp = NULL;
156
157 /*
158 * There are two arrays, one provided by the user (l1), and the
159 * temporary one (l2). The data is sorted to the temporary stack,
160 * and then copied back. The speedup of using index to determine
161 * which stack the data is on and simply swapping stacks back and
162 * forth, thus avoiding the copy every iteration, turns out to not
163 * be any faster than the current implementation.
164 */
165 if (!(l2 = (const u_char **)malloc(sizeof(u_char *) * nmemb)))
166 return(-1);
167
168 /*
169 * Tr references a table of sort weights; multiple entries may
170 * map to the same weight; EOS char must have the lowest weight.
171 */
172 if (tab)
173 tr = tab;
174 else {
175 for (t1 = 0, t2 = endbyte; t1 < t2; ++t1)
176 ltab[t1] = t1 + 1;
177 ltab[t2] = 0;
178 for (t1 = endbyte + 1; t1 < NBUCKETS; ++t1)
179 ltab[t1] = t1;
180 tr = ltab;
181 }
182
183 /* First sort is entire stack */
184 bot = l1;
185 indx = 0;
186
187 for (;;) {
188 /* Clear bucket count array */
189 bzero((char *)c, sizeof(c));
190
191 /*
192 * Compute number of items that sort to the same bucket
193 * for this index.
194 */
195 for (p = bot, i = nmemb; --i >= 0;)
196 ++c[tr[(*p++)[indx]]];
197
198 /*
199 * Sum the number of characters into c, dividing the temp
200 * stack into the right number of buckets for this bucket,
201 * this index. C contains the cumulative total of keys
202 * before and included in this bucket, and will later be
203 * used as an index to the bucket. c[NBUCKETS] contains
204 * the total number of elements, for determining how many
205 * elements the last bucket contains. At the same time
206 * find the largest bucket so it gets pushed first.
207 */
208 for (i = max = t1 = 0, t2 = __rspartition; i <= NBUCKETS; ++i) {
209 if (c[i] > t2) {
210 t2 = c[i];
211 max = i;
212 }
213 t1 = c[i] += t1;
214 }
215
216 /*
217 * Partition the elements into buckets; c decrements through
218 * the bucket, and ends up pointing to the first element of
219 * the bucket.
220 */
221 for (i = nmemb; --i >= 0;) {
222 --p;
223 l2[--c[tr[(*p)[indx]]]] = *p;
224 }
225
226 /* Copy the partitioned elements back to user stack */
227 bcopy(l2, bot, nmemb * sizeof(u_char *));
228
229 ++indx;
230 /*
231 * Sort buckets as necessary; don't sort c[0], it's the
232 * EOS character bucket, and nothing can follow EOS.
233 */
234 for (i = max; i; --i) {
235 if ((nmemb = c[i + 1] - (t1 = c[i])) < 2)
236 continue;
237 p = bot + t1;
238 if (nmemb > __rspartition)
239 STACKPUSH
240 else
241 shellsort(p, indx, nmemb, tr);
242 }
243 for (i = max + 1; i < NBUCKETS; ++i) {
244 if ((nmemb = c[i + 1] - (t1 = c[i])) < 2)
245 continue;
246 p = bot + t1;
247 if (nmemb > __rspartition)
248 STACKPUSH
249 else
250 shellsort(p, indx, nmemb, tr);
251 }
252 /* Break out when stack is empty */
253 STACKPOP
254 }
255
256 free((char *)l2);
257 free((char *)stack);
258 return(0);
259 }
260
261 /*
262 * Shellsort (diminishing increment sort) from Data Structures and
263 * Algorithms, Aho, Hopcraft and Ullman, 1983 Edition, page 290;
264 * see also Knuth Vol. 3, page 84. The increments are selected from
265 * formula (8), page 95. Roughly O(N^3/2).
266 */
267 static void
268 shellsort(p, indx, nmemb, tr)
269 register u_char **p, *tr;
270 register int indx, nmemb;
271 {
272 register u_char ch, *s1, *s2;
273 register int incr, *incrp, t1, t2;
274
275 for (incrp = __rsshell_increments; incr = *incrp++;)
276 for (t1 = incr; t1 < nmemb; ++t1)
277 for (t2 = t1 - incr; t2 >= 0;) {
278 s1 = p[t2] + indx;
279 s2 = p[t2 + incr] + indx;
280 while ((ch = tr[*s1++]) == tr[*s2] && ch)
281 ++s2;
282 if (ch > tr[*s2]) {
283 s1 = p[t2];
284 p[t2] = p[t2 + incr];
285 p[t2 + incr] = s1;
286 t2 -= incr;
287 } else
288 break;
289 }
290 }
291