radixsort.c revision 1.12
1 /* $NetBSD: radixsort.c,v 1.12 1999/09/16 11:45:35 lukem Exp $ */ 2 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. All advertising materials mentioning features or use of this software 19 * must display the following acknowledgement: 20 * This product includes software developed by the University of 21 * California, Berkeley and its contributors. 22 * 4. Neither the name of the University nor the names of its contributors 23 * may be used to endorse or promote products derived from this software 24 * without specific prior written permission. 25 * 26 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 27 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 28 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 29 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 30 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 31 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 32 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 33 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 34 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 35 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 36 * SUCH DAMAGE. 37 */ 38 39 #include <sys/cdefs.h> 40 #if defined(LIBC_SCCS) && !defined(lint) 41 #if 0 42 static char sccsid[] = "@(#)radixsort.c 8.2 (Berkeley) 4/28/95"; 43 #else 44 __RCSID("$NetBSD: radixsort.c,v 1.12 1999/09/16 11:45:35 lukem Exp $"); 45 #endif 46 #endif /* LIBC_SCCS and not lint */ 47 48 /* 49 * Radixsort routines. 50 * 51 * Program r_sort_a() is unstable but uses O(logN) extra memory for a stack. 52 * Use radixsort(a, n, trace, endchar) for this case. 53 * 54 * For stable sorting (using N extra pointers) use sradixsort(), which calls 55 * r_sort_b(). 56 * 57 * For a description of this code, see D. McIlroy, P. McIlroy, K. Bostic, 58 * "Engineering Radix Sort". 59 */ 60 61 #include "namespace.h" 62 #include <sys/types.h> 63 64 #include <assert.h> 65 #include <errno.h> 66 #include <stdlib.h> 67 68 #ifdef __weak_alias 69 __weak_alias(radixsort,_radixsort); 70 __weak_alias(sradixsort,_sradixsort); 71 #endif 72 73 typedef struct { 74 const u_char **sa; 75 int sn, si; 76 } stack; 77 78 static __inline void simplesort 79 __P((const u_char **, int, int, const u_char *, u_int)); 80 static void r_sort_a __P((const u_char **, int, int, const u_char *, u_int)); 81 static void r_sort_b __P((const u_char **, 82 const u_char **, int, int, const u_char *, u_int)); 83 84 #define THRESHOLD 20 /* Divert to simplesort(). */ 85 #define SIZE 512 /* Default stack size. */ 86 87 #define SETUP { \ 88 if (tab == NULL) { \ 89 tr = tr0; \ 90 for (c = 0; c < endch; c++) \ 91 tr0[c] = c + 1; \ 92 tr0[c] = 0; \ 93 for (c++; c < 256; c++) \ 94 tr0[c] = c; \ 95 endch = 0; \ 96 } else { \ 97 endch = tab[endch]; \ 98 tr = tab; \ 99 if (endch != 0 && endch != 255) { \ 100 errno = EINVAL; \ 101 return (-1); \ 102 } \ 103 } \ 104 } 105 106 int 107 radixsort(a, n, tab, endch) 108 const u_char **a, *tab; 109 int n; 110 u_int endch; 111 { 112 const u_char *tr; 113 int c; 114 u_char tr0[256]; 115 116 _DIAGASSERT(a != NULL); 117 _DIAGASSERT(tab != NULL); 118 #ifdef _DIAGNOSTIC 119 if (a == NULL || tab == NULL) { 120 errno = EFAULT; 121 return (-1); 122 } 123 #endif 124 125 SETUP; 126 r_sort_a(a, n, 0, tr, endch); 127 return (0); 128 } 129 130 int 131 sradixsort(a, n, tab, endch) 132 const u_char **a, *tab; 133 int n; 134 u_int endch; 135 { 136 const u_char *tr, **ta; 137 int c; 138 u_char tr0[256]; 139 140 _DIAGASSERT(a != NULL); 141 _DIAGASSERT(tab != NULL); 142 if (a == NULL || tab == NULL) { 143 errno = EFAULT; 144 return (-1); 145 } 146 147 SETUP; 148 if (n < THRESHOLD) 149 simplesort(a, n, 0, tr, endch); 150 else { 151 if ((ta = malloc(n * sizeof(a))) == NULL) 152 return (-1); 153 r_sort_b(a, ta, n, 0, tr, endch); 154 free(ta); 155 } 156 return (0); 157 } 158 159 #define empty(s) (s >= sp) 160 #define pop(a, n, i) a = (--sp)->sa, n = sp->sn, i = sp->si 161 #define push(a, n, i) sp->sa = a, sp->sn = n, (sp++)->si = i 162 #define swap(a, b, t) t = a, a = b, b = t 163 164 /* Unstable, in-place sort. */ 165 static void 166 r_sort_a(a, n, i, tr, endch) 167 const u_char **a; 168 int n, i; 169 const u_char *tr; 170 u_int endch; 171 { 172 static int count[256], nc, bmin; 173 int c; 174 const u_char **ak, *r; 175 stack s[SIZE], *sp, *sp0, *sp1, temp; 176 int *cp, bigc; 177 const u_char **an, *t, **aj, **top[256]; 178 179 _DIAGASSERT(a != NULL); 180 _DIAGASSERT(tr != NULL); 181 182 /* Set up stack. */ 183 sp = s; 184 push(a, n, i); 185 while (!empty(s)) { 186 pop(a, n, i); 187 if (n < THRESHOLD) { 188 simplesort(a, n, i, tr, endch); 189 continue; 190 } 191 an = a + n; 192 193 /* Make character histogram. */ 194 if (nc == 0) { 195 bmin = 255; /* First occupied bin, excluding eos. */ 196 for (ak = a; ak < an;) { 197 c = tr[(*ak++)[i]]; 198 if (++count[c] == 1 && c != endch) { 199 if (c < bmin) 200 bmin = c; 201 nc++; 202 } 203 } 204 if (sp + nc > s + SIZE) { /* Get more stack. */ 205 r_sort_a(a, n, i, tr, endch); 206 continue; 207 } 208 } 209 210 /* 211 * Set top[]; push incompletely sorted bins onto stack. 212 * top[] = pointers to last out-of-place element in bins. 213 * count[] = counts of elements in bins. 214 * Before permuting: top[c-1] + count[c] = top[c]; 215 * during deal: top[c] counts down to top[c-1]. 216 */ 217 sp0 = sp1 = sp; /* Stack position of biggest bin. */ 218 bigc = 2; /* Size of biggest bin. */ 219 if (endch == 0) /* Special case: set top[eos]. */ 220 top[0] = ak = a + count[0]; 221 else { 222 ak = a; 223 top[255] = an; 224 } 225 for (cp = count + bmin; nc > 0; cp++) { 226 while (*cp == 0) /* Find next non-empty pile. */ 227 cp++; 228 if (*cp > 1) { 229 if (*cp > bigc) { 230 bigc = *cp; 231 sp1 = sp; 232 } 233 push(ak, *cp, i+1); 234 } 235 top[cp-count] = ak += *cp; 236 nc--; 237 } 238 swap(*sp0, *sp1, temp); /* Play it safe -- biggest bin last. */ 239 240 /* 241 * Permute misplacements home. Already home: everything 242 * before aj, and in bin[c], items from top[c] on. 243 * Inner loop: 244 * r = next element to put in place; 245 * ak = top[r[i]] = location to put the next element. 246 * aj = bottom of 1st disordered bin. 247 * Outer loop: 248 * Once the 1st disordered bin is done, ie. aj >= ak, 249 * aj<-aj + count[c] connects the bins in a linked list; 250 * reset count[c]. 251 */ 252 for (aj = a; aj < an; *aj = r, aj += count[c], count[c] = 0) 253 for (r = *aj; aj < (ak = --top[c = tr[r[i]]]);) 254 swap(*ak, r, t); 255 } 256 } 257 258 /* Stable sort, requiring additional memory. */ 259 static void 260 r_sort_b(a, ta, n, i, tr, endch) 261 const u_char **a, **ta; 262 int n, i; 263 const u_char *tr; 264 u_int endch; 265 { 266 static int count[256], nc, bmin; 267 int c; 268 const u_char **ak, **ai; 269 stack s[512], *sp, *sp0, *sp1, temp; 270 const u_char **top[256]; 271 int *cp, bigc; 272 273 _DIAGASSERT(a != NULL); 274 _DIAGASSERT(ta != NULL); 275 _DIAGASSERT(tr != NULL); 276 277 sp = s; 278 push(a, n, i); 279 while (!empty(s)) { 280 pop(a, n, i); 281 if (n < THRESHOLD) { 282 simplesort(a, n, i, tr, endch); 283 continue; 284 } 285 286 if (nc == 0) { 287 bmin = 255; 288 for (ak = a + n; --ak >= a;) { 289 c = tr[(*ak)[i]]; 290 if (++count[c] == 1 && c != endch) { 291 if (c < bmin) 292 bmin = c; 293 nc++; 294 } 295 } 296 if (sp + nc > s + SIZE) { 297 r_sort_b(a, ta, n, i, tr, endch); 298 continue; 299 } 300 } 301 302 sp0 = sp1 = sp; 303 bigc = 2; 304 if (endch == 0) { 305 top[0] = ak = a + count[0]; 306 count[0] = 0; 307 } else { 308 ak = a; 309 top[255] = a + n; 310 count[255] = 0; 311 } 312 for (cp = count + bmin; nc > 0; cp++) { 313 while (*cp == 0) 314 cp++; 315 if ((c = *cp) > 1) { 316 if (c > bigc) { 317 bigc = c; 318 sp1 = sp; 319 } 320 push(ak, c, i+1); 321 } 322 top[cp-count] = ak += c; 323 *cp = 0; /* Reset count[]. */ 324 nc--; 325 } 326 swap(*sp0, *sp1, temp); 327 328 for (ak = ta + n, ai = a+n; ak > ta;) /* Copy to temp. */ 329 *--ak = *--ai; 330 for (ak = ta+n; --ak >= ta;) /* Deal to piles. */ 331 *--top[tr[(*ak)[i]]] = *ak; 332 } 333 } 334 335 static __inline void 336 simplesort(a, n, b, tr, endch) /* insertion sort */ 337 const u_char **a; 338 int n, b; 339 const u_char *tr; 340 u_int endch; 341 { 342 u_char ch; 343 const u_char **ak, **ai, *s, *t; 344 345 _DIAGASSERT(a != NULL); 346 _DIAGASSERT(tr != NULL); 347 348 for (ak = a+1; --n >= 1; ak++) 349 for (ai = ak; ai > a; ai--) { 350 for (s = ai[0] + b, t = ai[-1] + b; 351 (ch = tr[*s]) != endch; s++, t++) 352 if (ch != tr[*t]) 353 break; 354 if (ch >= tr[*t]) 355 break; 356 swap(ai[0], ai[-1], s); 357 } 358 } 359
Indexes created Sat Sep 20 22:09:52 GMT 2025