1 /*- 2 * Copyright (c) 1990, 1993 3 * The Regents of the University of California. All rights reserved. 4 * 5 * This code is derived from software contributed to Berkeley by 6 * Peter McIlroy and by Dan Bernstein at New York University, 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. All advertising materials mentioning features or use of this software 17 * must display the following acknowledgement: 18 * This product includes software developed by the University of 19 * California, Berkeley and its contributors. 20 * 4. Neither the name of the University nor the names of its contributors 21 * may be used to endorse or promote products derived from this software 22 * without specific prior written permission. 23 * 24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 34 * SUCH DAMAGE. 35 */ 36 37 #if defined(LIBC_SCCS) && !defined(lint) 38 static char *rcsid = "$OpenBSD: radixsort.c,v 1.5 2002/02/17 19:42:24 millert Exp $"; 39 #endif /* LIBC_SCCS and not lint */ 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 <errno.h> 57 58 typedef struct { 59 const u_char **sa; 60 int sn, si; 61 } stack; 62 63 static __inline void simplesort 64 (const u_char **, int, int, const u_char *, u_int); 65 static void r_sort_a(const u_char **, int, int, const u_char *, u_int); 66 static void r_sort_b(const u_char **, 67 const u_char **, int, int, const u_char *, u_int); 68 69 #define THRESHOLD 20 /* Divert to simplesort(). */ 70 #define SIZE 512 /* Default stack size. */ 71 72 #define SETUP { \ 73 if (tab == NULL) { \ 74 tr = tr0; \ 75 for (c = 0; c < endch; c++) \ 76 tr0[c] = c + 1; \ 77 tr0[c] = 0; \ 78 for (c++; c < 256; c++) \ 79 tr0[c] = c; \ 80 endch = 0; \ 81 } else { \ 82 endch = tab[endch]; \ 83 tr = tab; \ 84 if (endch != 0 && endch != 255) { \ 85 errno = EINVAL; \ 86 return (-1); \ 87 } \ 88 } \ 89 } 90 91 int 92 radixsort(a, n, tab, endch) 93 const u_char **a, *tab; 94 int n; 95 u_int endch; 96 { 97 const u_char *tr; 98 int c; 99 u_char tr0[256]; 100 101 SETUP; 102 r_sort_a(a, n, 0, tr, endch); 103 return (0); 104 } 105 106 int 107 sradixsort(a, n, tab, endch) 108 const u_char **a, *tab; 109 int n; 110 u_int endch; 111 { 112 const u_char *tr, **ta; 113 int c; 114 u_char tr0[256]; 115 116 SETUP; 117 if (n < THRESHOLD) 118 simplesort(a, n, 0, tr, endch); 119 else { 120 if ((ta = malloc(n * sizeof(a))) == NULL) 121 return (-1); 122 r_sort_b(a, ta, n, 0, tr, endch); 123 free(ta); 124 } 125 return (0); 126 } 127 128 #define empty(s) (s >= sp) 129 #define pop(a, n, i) a = (--sp)->sa, n = sp->sn, i = sp->si 130 #define push(a, n, i) sp->sa = a, sp->sn = n, (sp++)->si = i 131 #define swap(a, b, t) t = a, a = b, b = t 132 133 /* Unstable, in-place sort. */ 134 void 135 r_sort_a(a, n, i, tr, endch) 136 const u_char **a; 137 int n, i; 138 const u_char *tr; 139 u_int endch; 140 { 141 static int count[256], nc, bmin; 142 register int c; 143 register const u_char **ak, *r; 144 stack s[SIZE], *sp, *sp0, *sp1, temp; 145 int *cp, bigc; 146 const u_char **an, *t, **aj, **top[256]; 147 148 /* Set up stack. */ 149 sp = s; 150 push(a, n, i); 151 while (!empty(s)) { 152 pop(a, n, i); 153 if (n < THRESHOLD) { 154 simplesort(a, n, i, tr, endch); 155 continue; 156 } 157 an = a + n; 158 159 /* Make character histogram. */ 160 if (nc == 0) { 161 bmin = 255; /* First occupied bin, excluding eos. */ 162 for (ak = a; ak < an;) { 163 c = tr[(*ak++)[i]]; 164 if (++count[c] == 1 && c != endch) { 165 if (c < bmin) 166 bmin = c; 167 nc++; 168 } 169 } 170 if (sp + nc > s + SIZE) { /* Get more stack. */ 171 r_sort_a(a, n, i, tr, endch); 172 continue; 173 } 174 } 175 176 /* 177 * Set top[]; push incompletely sorted bins onto stack. 178 * top[] = pointers to last out-of-place element in bins. 179 * count[] = counts of elements in bins. 180 * Before permuting: top[c-1] + count[c] = top[c]; 181 * during deal: top[c] counts down to top[c-1]. 182 */ 183 sp0 = sp1 = sp; /* Stack position of biggest bin. */ 184 bigc = 2; /* Size of biggest bin. */ 185 if (endch == 0) /* Special case: set top[eos]. */ 186 top[0] = ak = a + count[0]; 187 else { 188 ak = a; 189 top[255] = an; 190 } 191 for (cp = count + bmin; nc > 0; cp++) { 192 while (*cp == 0) /* Find next non-empty pile. */ 193 cp++; 194 if (*cp > 1) { 195 if (*cp > bigc) { 196 bigc = *cp; 197 sp1 = sp; 198 } 199 push(ak, *cp, i+1); 200 } 201 top[cp-count] = ak += *cp; 202 nc--; 203 } 204 swap(*sp0, *sp1, temp); /* Play it safe -- biggest bin last. */ 205 206 /* 207 * Permute misplacements home. Already home: everything 208 * before aj, and in bin[c], items from top[c] on. 209 * Inner loop: 210 * r = next element to put in place; 211 * ak = top[r[i]] = location to put the next element. 212 * aj = bottom of 1st disordered bin. 213 * Outer loop: 214 * Once the 1st disordered bin is done, ie. aj >= ak, 215 * aj<-aj + count[c] connects the bins in a linked list; 216 * reset count[c]. 217 */ 218 for (aj = a; aj < an; *aj = r, aj += count[c], count[c] = 0) 219 for (r = *aj; aj < (ak = --top[c = tr[r[i]]]);) 220 swap(*ak, r, t); 221 } 222 } 223 224 /* Stable sort, requiring additional memory. */ 225 void 226 r_sort_b(a, ta, n, i, tr, endch) 227 const u_char **a, **ta; 228 int n, i; 229 const u_char *tr; 230 u_int endch; 231 { 232 static int count[256], nc, bmin; 233 register int c; 234 register const u_char **ak, **ai; 235 stack s[512], *sp, *sp0, *sp1, temp; 236 const u_char **top[256]; 237 int *cp, bigc; 238 239 sp = s; 240 push(a, n, i); 241 while (!empty(s)) { 242 pop(a, n, i); 243 if (n < THRESHOLD) { 244 simplesort(a, n, i, tr, endch); 245 continue; 246 } 247 248 if (nc == 0) { 249 bmin = 255; 250 for (ak = a + n; --ak >= a;) { 251 c = tr[(*ak)[i]]; 252 if (++count[c] == 1 && c != endch) { 253 if (c < bmin) 254 bmin = c; 255 nc++; 256 } 257 } 258 if (sp + nc > s + SIZE) { 259 r_sort_b(a, ta, n, i, tr, endch); 260 continue; 261 } 262 } 263 264 sp0 = sp1 = sp; 265 bigc = 2; 266 if (endch == 0) { 267 top[0] = ak = a + count[0]; 268 count[0] = 0; 269 } else { 270 ak = a; 271 top[255] = a + n; 272 count[255] = 0; 273 } 274 for (cp = count + bmin; nc > 0; cp++) { 275 while (*cp == 0) 276 cp++; 277 if ((c = *cp) > 1) { 278 if (c > bigc) { 279 bigc = c; 280 sp1 = sp; 281 } 282 push(ak, c, i+1); 283 } 284 top[cp-count] = ak += c; 285 *cp = 0; /* Reset count[]. */ 286 nc--; 287 } 288 swap(*sp0, *sp1, temp); 289 290 for (ak = ta + n, ai = a+n; ak > ta;) /* Copy to temp. */ 291 *--ak = *--ai; 292 for (ak = ta+n; --ak >= ta;) /* Deal to piles. */ 293 *--top[tr[(*ak)[i]]] = *ak; 294 } 295 } 296 297 static __inline void 298 simplesort(a, n, b, tr, endch) /* insertion sort */ 299 register const u_char **a; 300 int n, b; 301 register const u_char *tr; 302 u_int endch; 303 { 304 register u_char ch; 305 const u_char **ak, **ai, *s, *t; 306 307 for (ak = a+1; --n >= 1; ak++) 308 for (ai = ak; ai > a; ai--) { 309 for (s = ai[0] + b, t = ai[-1] + b; 310 (ch = tr[*s]) != endch; s++, t++) 311 if (ch != tr[*t]) 312 break; 313 if (ch >= tr[*t]) 314 break; 315 swap(ai[0], ai[-1], s); 316 } 317 } 318