1 /* 2 * Copyright (c) 1989 The Regents of the University of California. 3 * All rights reserved. 4 * 5 * This code is derived from software contributed to Berkeley by 6 * Michael Rendell of Memorial University of Newfoundland. 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 #ifndef lint 38 char copyright[] = 39 "@(#) Copyright (c) 1989 The Regents of the University of California.\n\ 40 All rights reserved.\n"; 41 #endif /* not lint */ 42 43 #ifndef lint 44 /*static char sccsid[] = "from: @(#)tsort.c 5.3 (Berkeley) 6/1/90";*/ 45 static char rcsid[] = "$Id: tsort.c,v 1.2 1993/08/01 18:04:16 mycroft Exp $"; 46 #endif /* not lint */ 47 48 #include <sys/types.h> 49 #include <errno.h> 50 #include <stdio.h> 51 #include <ctype.h> 52 #include <string.h> 53 54 /* 55 * Topological sort. Input is a list of pairs of strings seperated by 56 * white space (spaces, tabs, and/or newlines); strings are written to 57 * standard output in sorted order, one per line. 58 * 59 * usage: 60 * tsort [inputfile] 61 * If no input file is specified, standard input is read. 62 * 63 * Should be compatable with AT&T tsort HOWEVER the output is not identical 64 * (i.e. for most graphs there is more than one sorted order, and this tsort 65 * usually generates a different one then the AT&T tsort). Also, cycle 66 * reporting seems to be more accurate in this version (the AT&T tsort 67 * sometimes says a node is in a cycle when it isn't). 68 * 69 * Michael Rendell, michael@stretch.cs.mun.ca - Feb 26, '90 70 */ 71 #define HASHSIZE 53 /* doesn't need to be big */ 72 #define NF_MARK 0x1 /* marker for cycle detection */ 73 #define NF_ACYCLIC 0x2 /* this node is cycle free */ 74 75 typedef struct node_str NODE; 76 77 struct node_str { 78 char *n_name; /* name of this node */ 79 NODE **n_prevp; /* pointer to previous node's n_next */ 80 NODE *n_next; /* next node in graph */ 81 NODE *n_hash; /* next node in hash table */ 82 int n_narcs; /* number of arcs in n_arcs[] */ 83 int n_arcsize; /* size of n_arcs[] array */ 84 NODE **n_arcs; /* array of arcs to other nodes */ 85 int n_refcnt; /* # of arcs pointing to this node */ 86 int n_flags; /* NF_* */ 87 }; 88 89 typedef struct _buf { 90 char *b_buf; 91 int b_bsize; 92 } BUF; 93 94 NODE *add_node(), *find_node(); 95 void add_arc(), no_memory(), remove_node(), tsort(); 96 char *grow_buf(), *malloc(); 97 98 extern int errno; 99 NODE *graph; 100 NODE *hashtable[HASHSIZE]; 101 NODE **cycle_buf; 102 NODE **longest_cycle; 103 104 main(argc, argv) 105 int argc; 106 char **argv; 107 { 108 register BUF *b; 109 register int c, n; 110 FILE *fp; 111 int bsize, nused; 112 BUF bufs[2]; 113 114 if (argc < 2) 115 fp = stdin; 116 else if (argc == 2) { 117 (void)fprintf(stderr, "usage: tsort [ inputfile ]\n"); 118 exit(1); 119 } else if (!(fp = fopen(argv[1], "r"))) { 120 (void)fprintf(stderr, "tsort: %s.\n", strerror(errno)); 121 exit(1); 122 } 123 124 for (b = bufs, n = 2; --n >= 0; b++) 125 b->b_buf = grow_buf((char *)NULL, b->b_bsize = 1024); 126 127 /* parse input and build the graph */ 128 for (n = 0, c = getc(fp);;) { 129 while (c != EOF && isspace(c)) 130 c = getc(fp); 131 if (c == EOF) 132 break; 133 134 nused = 0; 135 b = &bufs[n]; 136 bsize = b->b_bsize; 137 do { 138 b->b_buf[nused++] = c; 139 if (nused == bsize) { 140 bsize *= 2; 141 b->b_buf = grow_buf(b->b_buf, bsize); 142 } 143 c = getc(fp); 144 } while (c != EOF && !isspace(c)); 145 146 b->b_buf[nused] = '\0'; 147 b->b_bsize = bsize; 148 if (n) 149 add_arc(bufs[0].b_buf, bufs[1].b_buf); 150 n = !n; 151 } 152 (void)fclose(fp); 153 if (n) { 154 (void)fprintf(stderr, "tsort: odd data count.\n"); 155 exit(1); 156 } 157 158 /* do the sort */ 159 tsort(); 160 exit(0); 161 } 162 163 /* double the size of oldbuf and return a pointer to the new buffer. */ 164 char * 165 grow_buf(bp, size) 166 char *bp; 167 int size; 168 { 169 char *realloc(); 170 171 if (!(bp = realloc(bp, (u_int)size))) 172 no_memory(); 173 return(bp); 174 } 175 176 /* 177 * add an arc from node s1 to node s2 in the graph. If s1 or s2 are not in 178 * the graph, then add them. 179 */ 180 void 181 add_arc(s1, s2) 182 char *s1, *s2; 183 { 184 register NODE *n1; 185 NODE *n2; 186 int bsize; 187 188 n1 = find_node(s1); 189 if (!n1) 190 n1 = add_node(s1); 191 192 if (!strcmp(s1, s2)) 193 return; 194 195 n2 = find_node(s2); 196 if (!n2) 197 n2 = add_node(s2); 198 199 /* 200 * could check to see if this arc is here already, but it isn't 201 * worth the bother -- there usually isn't and it doesn't hurt if 202 * there is (I think :-). 203 */ 204 if (n1->n_narcs == n1->n_arcsize) { 205 if (!n1->n_arcsize) 206 n1->n_arcsize = 10; 207 bsize = n1->n_arcsize * sizeof(*n1->n_arcs) * 2; 208 n1->n_arcs = (NODE **)grow_buf((char *)n1->n_arcs, bsize); 209 n1->n_arcsize = bsize / sizeof(*n1->n_arcs); 210 } 211 n1->n_arcs[n1->n_narcs++] = n2; 212 ++n2->n_refcnt; 213 } 214 215 hash_string(s) 216 char *s; 217 { 218 register int hash, i; 219 220 for (hash = 0, i = 1; *s; s++, i++) 221 hash += *s * i; 222 return(hash % HASHSIZE); 223 } 224 225 /* 226 * find a node in the graph and return a pointer to it - returns null if not 227 * found. 228 */ 229 NODE * 230 find_node(name) 231 char *name; 232 { 233 register NODE *n; 234 235 for (n = hashtable[hash_string(name)]; n; n = n->n_hash) 236 if (!strcmp(n->n_name, name)) 237 return(n); 238 return((NODE *)NULL); 239 } 240 241 /* Add a node to the graph and return a pointer to it. */ 242 NODE * 243 add_node(name) 244 char *name; 245 { 246 register NODE *n; 247 int hash; 248 249 if (!(n = (NODE *)malloc(sizeof(NODE))) || !(n->n_name = strdup(name))) 250 no_memory(); 251 252 n->n_narcs = 0; 253 n->n_arcsize = 0; 254 n->n_arcs = (NODE **)NULL; 255 n->n_refcnt = 0; 256 n->n_flags = 0; 257 258 /* add to linked list */ 259 if (n->n_next = graph) 260 graph->n_prevp = &n->n_next; 261 n->n_prevp = &graph; 262 graph = n; 263 264 /* add to hash table */ 265 hash = hash_string(name); 266 n->n_hash = hashtable[hash]; 267 hashtable[hash] = n; 268 return(n); 269 } 270 271 /* do topological sort on graph */ 272 void 273 tsort() 274 { 275 register NODE *n, *next; 276 register int cnt; 277 278 while (graph) { 279 /* 280 * keep getting rid of simple cases until there are none left, 281 * if there are any nodes still in the graph, then there is 282 * a cycle in it. 283 */ 284 do { 285 for (cnt = 0, n = graph; n; n = next) { 286 next = n->n_next; 287 if (n->n_refcnt == 0) { 288 remove_node(n); 289 ++cnt; 290 } 291 } 292 } while (graph && cnt); 293 294 if (!graph) 295 break; 296 297 if (!cycle_buf) { 298 /* 299 * allocate space for two cycle logs - one to be used 300 * as scratch space, the other to save the longest 301 * cycle. 302 */ 303 for (cnt = 0, n = graph; n; n = n->n_next) 304 ++cnt; 305 cycle_buf = 306 (NODE **)malloc((u_int)sizeof(NODE *) * cnt); 307 longest_cycle = 308 (NODE **)malloc((u_int)sizeof(NODE *) * cnt); 309 if (!cycle_buf || !longest_cycle) 310 no_memory(); 311 } 312 for (n = graph; n; n = n->n_next) 313 if (!(n->n_flags & NF_ACYCLIC)) { 314 if (cnt = find_cycle(n, n, 0, 0)) { 315 register int i; 316 317 (void)fprintf(stderr, 318 "tsort: cycle in data.\n"); 319 for (i = 0; i < cnt; i++) 320 (void)fprintf(stderr, 321 "tsort: %s.\n", longest_cycle[i]->n_name); 322 remove_node(n); 323 break; 324 } else 325 /* to avoid further checks */ 326 n->n_flags = NF_ACYCLIC; 327 } 328 329 if (!n) { 330 (void)fprintf(stderr, 331 "tsort: internal error -- could not find cycle.\n"); 332 exit(1); 333 } 334 } 335 } 336 337 /* print node and remove from graph (does not actually free node) */ 338 void 339 remove_node(n) 340 register NODE *n; 341 { 342 register NODE **np; 343 register int i; 344 345 (void)printf("%s\n", n->n_name); 346 for (np = n->n_arcs, i = n->n_narcs; --i >= 0; np++) 347 --(*np)->n_refcnt; 348 n->n_narcs = 0; 349 *n->n_prevp = n->n_next; 350 if (n->n_next) 351 n->n_next->n_prevp = n->n_prevp; 352 } 353 354 /* look for the longest cycle from node from to node to. */ 355 find_cycle(from, to, longest_len, depth) 356 NODE *from, *to; 357 int depth, longest_len; 358 { 359 register NODE **np; 360 register int i, len; 361 362 /* 363 * avoid infinite loops and ignore portions of the graph known 364 * to be acyclic 365 */ 366 if (from->n_flags & (NF_MARK|NF_ACYCLIC)) 367 return(0); 368 from->n_flags = NF_MARK; 369 370 for (np = from->n_arcs, i = from->n_narcs; --i >= 0; np++) { 371 cycle_buf[depth] = *np; 372 if (*np == to) { 373 if (depth + 1 > longest_len) { 374 longest_len = depth + 1; 375 (void)memcpy((char *)longest_cycle, 376 (char *)cycle_buf, 377 longest_len * sizeof(NODE *)); 378 } 379 } else { 380 len = find_cycle(*np, to, longest_len, depth + 1); 381 if (len > longest_len) 382 longest_len = len; 383 } 384 } 385 from->n_flags &= ~NF_MARK; 386 return(longest_len); 387 } 388 389 void 390 no_memory() 391 { 392 (void)fprintf(stderr, "tsort: %s.\n", strerror(ENOMEM)); 393 exit(1); 394 } 395