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 #include <sys/cdefs.h> 39 __FBSDID("$FreeBSD$"); 40 41 /* 42 * Radixsort routines. 43 * 44 * Program r_sort_a() is unstable but uses O(logN) extra memory for a stack. 45 * Use radixsort(a, n, trace, endchar) for this case. 46 * 47 * For stable sorting (using N extra pointers) use sradixsort(), which calls 48 * r_sort_b(). 49 * 50 * For a description of this code, see D. McIlroy, P. McIlroy, K. Bostic, 51 * "Engineering Radix Sort". 52 */ 53 54 #include <sys/types.h> 55 #include <stdlib.h> 56 #include <stddef.h> 57 #include <errno.h> 58 59 typedef struct { 60 const u_char **sa; 61 int sn, si; 62 } stack; 63 64 static inline void simplesort 65 (const u_char **, int, int, const u_char *, u_int); 66 static void r_sort_a(const u_char **, int, int, const u_char *, u_int); 67 static void r_sort_b(const u_char **, const u_char **, int, int, 68 const u_char *, u_int); 69 70 #define THRESHOLD 20 /* Divert to simplesort(). */ 71 #define SIZE 512 /* Default stack size. */ 72 73 #define SETUP { \ 74 if (tab == NULL) { \ 75 tr = tr0; \ 76 for (c = 0; c < endch; c++) \ 77 tr0[c] = c + 1; \ 78 tr0[c] = 0; \ 79 for (c++; c < 256; c++) \ 80 tr0[c] = c; \ 81 endch = 0; \ 82 } else { \ 83 endch = tab[endch]; \ 84 tr = tab; \ 85 if (endch != 0 && endch != 255) { \ 86 errno = EINVAL; \ 87 return (-1); \ 88 } \ 89 } \ 90 } 91 92 int 93 radixsort(const u_char **a, int n, const u_char *tab, u_int endch) 94 { 95 const u_char *tr; 96 int c; 97 u_char tr0[256]; 98 99 SETUP; 100 r_sort_a(a, n, 0, tr, endch); 101 return (0); 102 } 103 104 int 105 sradixsort(const u_char **a, int n, const u_char *tab, u_int endch) 106 { 107 const u_char *tr, **ta; 108 int c; 109 u_char tr0[256]; 110 111 SETUP; 112 if (n < THRESHOLD) 113 simplesort(a, n, 0, tr, endch); 114 else { 115 if ((ta = malloc(n * sizeof(a))) == NULL) 116 return (-1); 117 r_sort_b(a, ta, n, 0, tr, endch); 118 free(ta); 119 } 120 return (0); 121 } 122 123 #define empty(s) (s >= sp) 124 #define pop(a, n, i) a = (--sp)->sa, n = sp->sn, i = sp->si 125 #define push(a, n, i) sp->sa = a, sp->sn = n, (sp++)->si = i 126 #define swap(a, b, t) t = a, a = b, b = t 127 128 /* Unstable, in-place sort. */ 129 static void 130 r_sort_a(const u_char **a, int n, int i, const u_char *tr, u_int endch) 131 { 132 static int count[256], nc, bmin; 133 int c; 134 const u_char **ak, *r; 135 stack s[SIZE], *sp, *sp0, *sp1, temp; 136 int *cp, bigc; 137 const u_char **an, *t, **aj, **top[256]; 138 139 /* Set up stack. */ 140 sp = s; 141 push(a, n, i); 142 while (!empty(s)) { 143 pop(a, n, i); 144 if (n < THRESHOLD) { 145 simplesort(a, n, i, tr, endch); 146 continue; 147 } 148 an = a + n; 149 150 /* Make character histogram. */ 151 if (nc == 0) { 152 bmin = 255; /* First occupied bin, excluding eos. */ 153 for (ak = a; ak < an;) { 154 c = tr[(*ak++)[i]]; 155 if (++count[c] == 1 && c != endch) { 156 if (c < bmin) 157 bmin = c; 158 nc++; 159 } 160 } 161 if (sp + nc > s + SIZE) { /* Get more stack. */ 162 r_sort_a(a, n, i, tr, endch); 163 continue; 164 } 165 } 166 167 /* 168 * Special case: if all strings have the same 169 * character at position i, move on to the next 170 * character. 171 */ 172 if (nc == 1 && count[bmin] == n) { 173 push(a, n, i+1); 174 nc = count[bmin] = 0; 175 continue; 176 } 177 178 /* 179 * Set top[]; push incompletely sorted bins onto stack. 180 * top[] = pointers to last out-of-place element in bins. 181 * count[] = counts of elements in bins. 182 * Before permuting: top[c-1] + count[c] = top[c]; 183 * during deal: top[c] counts down to top[c-1]. 184 */ 185 sp0 = sp1 = sp; /* Stack position of biggest bin. */ 186 bigc = 2; /* Size of biggest bin. */ 187 if (endch == 0) /* Special case: set top[eos]. */ 188 top[0] = ak = a + count[0]; 189 else { 190 ak = a; 191 top[255] = an; 192 } 193 for (cp = count + bmin; nc > 0; cp++) { 194 while (*cp == 0) /* Find next non-empty pile. */ 195 cp++; 196 if (*cp > 1) { 197 if (*cp > bigc) { 198 bigc = *cp; 199 sp1 = sp; 200 } 201 push(ak, *cp, i+1); 202 } 203 top[cp-count] = ak += *cp; 204 nc--; 205 } 206 swap(*sp0, *sp1, temp); /* Play it safe -- biggest bin last. */ 207 208 /* 209 * Permute misplacements home. Already home: everything 210 * before aj, and in bin[c], items from top[c] on. 211 * Inner loop: 212 * r = next element to put in place; 213 * ak = top[r[i]] = location to put the next element. 214 * aj = bottom of 1st disordered bin. 215 * Outer loop: 216 * Once the 1st disordered bin is done, ie. aj >= ak, 217 * aj<-aj + count[c] connects the bins in a linked list; 218 * reset count[c]. 219 */ 220 for (aj = a; aj < an; *aj = r, aj += count[c], count[c] = 0) 221 for (r = *aj; aj < (ak = --top[c = tr[r[i]]]);) 222 swap(*ak, r, t); 223 } 224 } 225 226 /* Stable sort, requiring additional memory. */ 227 static void 228 r_sort_b(const u_char **a, const u_char **ta, int n, int i, const u_char *tr, 229 u_int endch) 230 { 231 static int count[256], nc, bmin; 232 int c; 233 const u_char **ak, **ai; 234 stack s[512], *sp, *sp0, *sp1, temp; 235 const u_char **top[256]; 236 int *cp, bigc; 237 238 sp = s; 239 push(a, n, i); 240 while (!empty(s)) { 241 pop(a, n, i); 242 if (n < THRESHOLD) { 243 simplesort(a, n, i, tr, endch); 244 continue; 245 } 246 247 if (nc == 0) { 248 bmin = 255; 249 for (ak = a + n; --ak >= a;) { 250 c = tr[(*ak)[i]]; 251 if (++count[c] == 1 && c != endch) { 252 if (c < bmin) 253 bmin = c; 254 nc++; 255 } 256 } 257 if (sp + nc > s + SIZE) { 258 r_sort_b(a, ta, n, i, tr, endch); 259 continue; 260 } 261 } 262 263 sp0 = sp1 = sp; 264 bigc = 2; 265 if (endch == 0) { 266 top[0] = ak = a + count[0]; 267 count[0] = 0; 268 } else { 269 ak = a; 270 top[255] = a + n; 271 count[255] = 0; 272 } 273 for (cp = count + bmin; nc > 0; cp++) { 274 while (*cp == 0) 275 cp++; 276 if ((c = *cp) > 1) { 277 if (c > bigc) { 278 bigc = c; 279 sp1 = sp; 280 } 281 push(ak, c, i+1); 282 } 283 top[cp-count] = ak += c; 284 *cp = 0; /* Reset count[]. */ 285 nc--; 286 } 287 swap(*sp0, *sp1, temp); 288 289 for (ak = ta + n, ai = a+n; ak > ta;) /* Copy to temp. */ 290 *--ak = *--ai; 291 for (ak = ta+n; --ak >= ta;) /* Deal to piles. */ 292 *--top[tr[(*ak)[i]]] = *ak; 293 } 294 } 295 296 /* insertion sort */ 297 static inline void 298 simplesort(const u_char **a, int n, int b, const u_char *tr, u_int endch) 299 { 300 u_char ch; 301 const u_char **ak, **ai, *s, *t; 302 303 for (ak = a+1; --n >= 1; ak++) 304 for (ai = ak; ai > a; ai--) { 305 for (s = ai[0] + b, t = ai[-1] + b; 306 (ch = tr[*s]) != endch; s++, t++) 307 if (ch != tr[*t]) 308 break; 309 if (ch >= tr[*t]) 310 break; 311 swap(ai[0], ai[-1], s); 312 } 313 } 314