1 /*-
2 * SPDX-License-Identifier: BSD-3-Clause
3 *
4 * Copyright (c) 1990, 1993
5 * The Regents of the University of California. All rights reserved.
6 *
7 * This code is derived from software contributed to Berkeley by
8 * Peter McIlroy and by Dan Bernstein at New York University,
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 * 3. Neither the name of the University nor the names of its contributors
19 * may be used to endorse or promote products derived from this software
20 * without specific prior written permission.
21 *
22 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32 * SUCH DAMAGE.
33 */
34
35 #if defined(LIBC_SCCS) && !defined(lint)
36 static char sccsid[] = "@(#)radixsort.c 8.2 (Berkeley) 4/28/95";
37 #endif /* LIBC_SCCS and not lint */
38 /*
39 * Radixsort routines.
40 *
41 * Program r_sort_a() is unstable but uses O(logN) extra memory for a stack.
42 * Use radixsort(a, n, trace, endchar) for this case.
43 *
44 * For stable sorting (using N extra pointers) use sradixsort(), which calls
45 * r_sort_b().
46 *
47 * For a description of this code, see D. McIlroy, P. McIlroy, K. Bostic,
48 * "Engineering Radix Sort".
49 */
50
51 #include <sys/types.h>
52 #include <stdlib.h>
53 #include <stddef.h>
54 #include <errno.h>
55
56 typedef struct {
57 const u_char **sa;
58 int sn, si;
59 } stack;
60
61 static inline void simplesort
62 (const u_char **, int, int, const u_char *, u_int);
63 static void r_sort_a(const u_char **, int, int, const u_char *, u_int);
64 static void r_sort_b(const u_char **, const u_char **, int, int,
65 const u_char *, u_int);
66
67 #define THRESHOLD 20 /* Divert to simplesort(). */
68 #define SIZE 512 /* Default stack size. */
69
70 #define SETUP { \
71 if (tab == NULL) { \
72 tr = tr0; \
73 for (c = 0; c < endch; c++) \
74 tr0[c] = c + 1; \
75 tr0[c] = 0; \
76 for (c++; c < 256; c++) \
77 tr0[c] = c; \
78 endch = 0; \
79 } else { \
80 endch = tab[endch]; \
81 tr = tab; \
82 if (endch != 0 && endch != 255) { \
83 errno = EINVAL; \
84 return (-1); \
85 } \
86 } \
87 }
88
89 int
radixsort(const u_char ** a,int n,const u_char * tab,u_int endch)90 radixsort(const u_char **a, int n, const u_char *tab, u_int endch)
91 {
92 const u_char *tr;
93 int c;
94 u_char tr0[256];
95
96 SETUP;
97 r_sort_a(a, n, 0, tr, endch);
98 return (0);
99 }
100
101 int
sradixsort(const u_char ** a,int n,const u_char * tab,u_int endch)102 sradixsort(const u_char **a, int n, const u_char *tab, u_int endch)
103 {
104 const u_char *tr, **ta;
105 int c;
106 u_char tr0[256];
107
108 SETUP;
109 if (n < THRESHOLD)
110 simplesort(a, n, 0, tr, endch);
111 else {
112 if ((ta = malloc(n * sizeof(a))) == NULL)
113 return (-1);
114 r_sort_b(a, ta, n, 0, tr, endch);
115 free(ta);
116 }
117 return (0);
118 }
119
120 #define empty(s) (s >= sp)
121 #define pop(a, n, i) a = (--sp)->sa, n = sp->sn, i = sp->si
122 #define push(a, n, i) sp->sa = a, sp->sn = n, (sp++)->si = i
123 #define swap(a, b, t) t = a, a = b, b = t
124
125 /* Unstable, in-place sort. */
126 static void
r_sort_a(const u_char ** a,int n,int i,const u_char * tr,u_int endch)127 r_sort_a(const u_char **a, int n, int i, const u_char *tr, u_int endch)
128 {
129 static int count[256], nc, bmin;
130 int c;
131 const u_char **ak, *r;
132 stack s[SIZE], *sp, *sp0, *sp1, temp;
133 int *cp, bigc;
134 const u_char **an, *t, **aj, **top[256];
135
136 /* Set up stack. */
137 sp = s;
138 push(a, n, i);
139 while (!empty(s)) {
140 pop(a, n, i);
141 if (n < THRESHOLD) {
142 simplesort(a, n, i, tr, endch);
143 continue;
144 }
145 an = a + n;
146
147 /* Make character histogram. */
148 if (nc == 0) {
149 bmin = 255; /* First occupied bin, excluding eos. */
150 for (ak = a; ak < an;) {
151 c = tr[(*ak++)[i]];
152 if (++count[c] == 1 && c != endch) {
153 if (c < bmin)
154 bmin = c;
155 nc++;
156 }
157 }
158 if (sp + nc > s + SIZE) { /* Get more stack. */
159 r_sort_a(a, n, i, tr, endch);
160 continue;
161 }
162 }
163
164 /*
165 * Special case: if all strings have the same
166 * character at position i, move on to the next
167 * character.
168 */
169 if (nc == 1 && count[bmin] == n) {
170 push(a, n, i+1);
171 nc = count[bmin] = 0;
172 continue;
173 }
174
175 /*
176 * Set top[]; push incompletely sorted bins onto stack.
177 * top[] = pointers to last out-of-place element in bins.
178 * count[] = counts of elements in bins.
179 * Before permuting: top[c-1] + count[c] = top[c];
180 * during deal: top[c] counts down to top[c-1].
181 */
182 sp0 = sp1 = sp; /* Stack position of biggest bin. */
183 bigc = 2; /* Size of biggest bin. */
184 if (endch == 0) /* Special case: set top[eos]. */
185 top[0] = ak = a + count[0];
186 else {
187 ak = a;
188 top[255] = an;
189 }
190 for (cp = count + bmin; nc > 0; cp++) {
191 while (*cp == 0) /* Find next non-empty pile. */
192 cp++;
193 if (*cp > 1) {
194 if (*cp > bigc) {
195 bigc = *cp;
196 sp1 = sp;
197 }
198 push(ak, *cp, i+1);
199 }
200 top[cp-count] = ak += *cp;
201 nc--;
202 }
203 swap(*sp0, *sp1, temp); /* Play it safe -- biggest bin last. */
204
205 /*
206 * Permute misplacements home. Already home: everything
207 * before aj, and in bin[c], items from top[c] on.
208 * Inner loop:
209 * r = next element to put in place;
210 * ak = top[r[i]] = location to put the next element.
211 * aj = bottom of 1st disordered bin.
212 * Outer loop:
213 * Once the 1st disordered bin is done, ie. aj >= ak,
214 * aj<-aj + count[c] connects the bins in a linked list;
215 * reset count[c].
216 */
217 for (aj = a; aj < an; *aj = r, aj += count[c], count[c] = 0)
218 for (r = *aj; aj < (ak = --top[c = tr[r[i]]]);)
219 swap(*ak, r, t);
220 }
221 }
222
223 /* Stable sort, requiring additional memory. */
224 static void
r_sort_b(const u_char ** a,const u_char ** ta,int n,int i,const u_char * tr,u_int endch)225 r_sort_b(const u_char **a, const u_char **ta, int n, int i, const u_char *tr,
226 u_int endch)
227 {
228 static int count[256], nc, bmin;
229 int c;
230 const u_char **ak, **ai;
231 stack s[512], *sp, *sp0, *sp1, temp;
232 const u_char **top[256];
233 int *cp, bigc;
234
235 sp = s;
236 push(a, n, i);
237 while (!empty(s)) {
238 pop(a, n, i);
239 if (n < THRESHOLD) {
240 simplesort(a, n, i, tr, endch);
241 continue;
242 }
243
244 if (nc == 0) {
245 bmin = 255;
246 for (ak = a + n; --ak >= a;) {
247 c = tr[(*ak)[i]];
248 if (++count[c] == 1 && c != endch) {
249 if (c < bmin)
250 bmin = c;
251 nc++;
252 }
253 }
254 if (sp + nc > s + SIZE) {
255 r_sort_b(a, ta, n, i, tr, endch);
256 continue;
257 }
258 }
259
260 sp0 = sp1 = sp;
261 bigc = 2;
262 if (endch == 0) {
263 top[0] = ak = a + count[0];
264 count[0] = 0;
265 } else {
266 ak = a;
267 top[255] = a + n;
268 count[255] = 0;
269 }
270 for (cp = count + bmin; nc > 0; cp++) {
271 while (*cp == 0)
272 cp++;
273 if ((c = *cp) > 1) {
274 if (c > bigc) {
275 bigc = c;
276 sp1 = sp;
277 }
278 push(ak, c, i+1);
279 }
280 top[cp-count] = ak += c;
281 *cp = 0; /* Reset count[]. */
282 nc--;
283 }
284 swap(*sp0, *sp1, temp);
285
286 for (ak = ta + n, ai = a+n; ak > ta;) /* Copy to temp. */
287 *--ak = *--ai;
288 for (ak = ta+n; --ak >= ta;) /* Deal to piles. */
289 *--top[tr[(*ak)[i]]] = *ak;
290 }
291 }
292
293 /* insertion sort */
294 static inline void
simplesort(const u_char ** a,int n,int b,const u_char * tr,u_int endch)295 simplesort(const u_char **a, int n, int b, const u_char *tr, u_int endch)
296 {
297 u_char ch;
298 const u_char **ak, **ai, *s, *t;
299
300 for (ak = a+1; --n >= 1; ak++)
301 for (ai = ak; ai > a; ai--) {
302 for (s = ai[0] + b, t = ai[-1] + b;
303 (ch = tr[*s]) != endch; s++, t++)
304 if (ch != tr[*t])
305 break;
306 if (ch >= tr[*t])
307 break;
308 swap(ai[0], ai[-1], s);
309 }
310 }
311