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. Neither the name of the University nor the names of its contributors 17 * may be used to endorse or promote products derived from this software 18 * without specific prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 23 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 30 * SUCH DAMAGE. 31 */ 32 33 #if defined(LIBC_SCCS) && !defined(lint) 34 static char sccsid[] = "@(#)radixsort.c 8.2 (Berkeley) 4/28/95"; 35 #endif /* LIBC_SCCS and not lint */ 36 #include <sys/cdefs.h> 37 __FBSDID("$FreeBSD$"); 38 39 /* 40 * Radixsort routines. 41 * 42 * Program r_sort_a() is unstable but uses O(logN) extra memory for a stack. 43 * Use radixsort(a, n, trace, endchar) for this case. 44 * 45 * For stable sorting (using N extra pointers) use sradixsort(), which calls 46 * r_sort_b(). 47 * 48 * For a description of this code, see D. McIlroy, P. McIlroy, K. Bostic, 49 * "Engineering Radix Sort". 50 */ 51 52 #include <sys/types.h> 53 #include <stdlib.h> 54 #include <stddef.h> 55 #include <errno.h> 56 57 typedef struct { 58 const u_char **sa; 59 int sn, si; 60 } stack; 61 62 static inline void simplesort 63 (const u_char **, int, int, const u_char *, u_int); 64 static void r_sort_a(const u_char **, int, int, const u_char *, u_int); 65 static void r_sort_b(const u_char **, const u_char **, int, int, 66 const u_char *, u_int); 67 68 #define THRESHOLD 20 /* Divert to simplesort(). */ 69 #define SIZE 512 /* Default stack size. */ 70 71 #define SETUP { \ 72 if (tab == NULL) { \ 73 tr = tr0; \ 74 for (c = 0; c < endch; c++) \ 75 tr0[c] = c + 1; \ 76 tr0[c] = 0; \ 77 for (c++; c < 256; c++) \ 78 tr0[c] = c; \ 79 endch = 0; \ 80 } else { \ 81 endch = tab[endch]; \ 82 tr = tab; \ 83 if (endch != 0 && endch != 255) { \ 84 errno = EINVAL; \ 85 return (-1); \ 86 } \ 87 } \ 88 } 89 90 int 91 radixsort(const u_char **a, int n, const u_char *tab, u_int endch) 92 { 93 const u_char *tr; 94 int c; 95 u_char tr0[256]; 96 97 SETUP; 98 r_sort_a(a, n, 0, tr, endch); 99 return (0); 100 } 101 102 int 103 sradixsort(const u_char **a, int n, const u_char *tab, u_int endch) 104 { 105 const u_char *tr, **ta; 106 int c; 107 u_char tr0[256]; 108 109 SETUP; 110 if (n < THRESHOLD) 111 simplesort(a, n, 0, tr, endch); 112 else { 113 if ((ta = malloc(n * sizeof(a))) == NULL) 114 return (-1); 115 r_sort_b(a, ta, n, 0, tr, endch); 116 free(ta); 117 } 118 return (0); 119 } 120 121 #define empty(s) (s >= sp) 122 #define pop(a, n, i) a = (--sp)->sa, n = sp->sn, i = sp->si 123 #define push(a, n, i) sp->sa = a, sp->sn = n, (sp++)->si = i 124 #define swap(a, b, t) t = a, a = b, b = t 125 126 /* Unstable, in-place sort. */ 127 static void 128 r_sort_a(const u_char **a, int n, int i, const u_char *tr, u_int endch) 129 { 130 static int count[256], nc, bmin; 131 int c; 132 const u_char **ak, *r; 133 stack s[SIZE], *sp, *sp0, *sp1, temp; 134 int *cp, bigc; 135 const u_char **an, *t, **aj, **top[256]; 136 137 /* Set up stack. */ 138 sp = s; 139 push(a, n, i); 140 while (!empty(s)) { 141 pop(a, n, i); 142 if (n < THRESHOLD) { 143 simplesort(a, n, i, tr, endch); 144 continue; 145 } 146 an = a + n; 147 148 /* Make character histogram. */ 149 if (nc == 0) { 150 bmin = 255; /* First occupied bin, excluding eos. */ 151 for (ak = a; ak < an;) { 152 c = tr[(*ak++)[i]]; 153 if (++count[c] == 1 && c != endch) { 154 if (c < bmin) 155 bmin = c; 156 nc++; 157 } 158 } 159 if (sp + nc > s + SIZE) { /* Get more stack. */ 160 r_sort_a(a, n, i, tr, endch); 161 continue; 162 } 163 } 164 165 /* 166 * Special case: if all strings have the same 167 * character at position i, move on to the next 168 * character. 169 */ 170 if (nc == 1 && count[bmin] == n) { 171 push(a, n, i+1); 172 nc = count[bmin] = 0; 173 continue; 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 static void 226 r_sort_b(const u_char **a, const u_char **ta, int n, int i, const u_char *tr, 227 u_int endch) 228 { 229 static int count[256], nc, bmin; 230 int c; 231 const u_char **ak, **ai; 232 stack s[512], *sp, *sp0, *sp1, temp; 233 const u_char **top[256]; 234 int *cp, bigc; 235 236 sp = s; 237 push(a, n, i); 238 while (!empty(s)) { 239 pop(a, n, i); 240 if (n < THRESHOLD) { 241 simplesort(a, n, i, tr, endch); 242 continue; 243 } 244 245 if (nc == 0) { 246 bmin = 255; 247 for (ak = a + n; --ak >= a;) { 248 c = tr[(*ak)[i]]; 249 if (++count[c] == 1 && c != endch) { 250 if (c < bmin) 251 bmin = c; 252 nc++; 253 } 254 } 255 if (sp + nc > s + SIZE) { 256 r_sort_b(a, ta, n, i, tr, endch); 257 continue; 258 } 259 } 260 261 sp0 = sp1 = sp; 262 bigc = 2; 263 if (endch == 0) { 264 top[0] = ak = a + count[0]; 265 count[0] = 0; 266 } else { 267 ak = a; 268 top[255] = a + n; 269 count[255] = 0; 270 } 271 for (cp = count + bmin; nc > 0; cp++) { 272 while (*cp == 0) 273 cp++; 274 if ((c = *cp) > 1) { 275 if (c > bigc) { 276 bigc = c; 277 sp1 = sp; 278 } 279 push(ak, c, i+1); 280 } 281 top[cp-count] = ak += c; 282 *cp = 0; /* Reset count[]. */ 283 nc--; 284 } 285 swap(*sp0, *sp1, temp); 286 287 for (ak = ta + n, ai = a+n; ak > ta;) /* Copy to temp. */ 288 *--ak = *--ai; 289 for (ak = ta+n; --ak >= ta;) /* Deal to piles. */ 290 *--top[tr[(*ak)[i]]] = *ak; 291 } 292 } 293 294 /* insertion sort */ 295 static inline void 296 simplesort(const u_char **a, int n, int b, const u_char *tr, u_int endch) 297 { 298 u_char ch; 299 const u_char **ak, **ai, *s, *t; 300 301 for (ak = a+1; --n >= 1; ak++) 302 for (ai = ak; ai > a; ai--) { 303 for (s = ai[0] + b, t = ai[-1] + b; 304 (ch = tr[*s]) != endch; s++, t++) 305 if (ch != tr[*t]) 306 break; 307 if (ch >= tr[*t]) 308 break; 309 swap(ai[0], ai[-1], s); 310 } 311 } 312