1 /* 2 * Copyright (c) 1989, 1993, 1994 3 * The Regents of the University of California. 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 * $Id: tsort.c,v 1.3 1996/06/10 16:12:43 phk Exp $ 37 */ 38 39 #ifndef lint 40 static char copyright[] = 41 "@(#) Copyright (c) 1989, 1993, 1994\n\ 42 The Regents of the University of California. All rights reserved.\n"; 43 #endif /* not lint */ 44 45 #ifndef lint 46 static char sccsid[] = "@(#)tsort.c 8.3 (Berkeley) 5/4/95"; 47 #endif /* not lint */ 48 49 #include <sys/types.h> 50 51 #include <ctype.h> 52 #include <db.h> 53 #include <err.h> 54 #include <errno.h> 55 #include <fcntl.h> 56 #include <stdio.h> 57 #include <stdlib.h> 58 #include <string.h> 59 60 /* 61 * Topological sort. Input is a list of pairs of strings separated by 62 * white space (spaces, tabs, and/or newlines); strings are written to 63 * standard output in sorted order, one per line. 64 * 65 * usage: 66 * tsort [-dlq] [inputfile] 67 * If no input file is specified, standard input is read. 68 * 69 * Should be compatable with AT&T tsort HOWEVER the output is not identical 70 * (i.e. for most graphs there is more than one sorted order, and this tsort 71 * usually generates a different one then the AT&T tsort). Also, cycle 72 * reporting seems to be more accurate in this version (the AT&T tsort 73 * sometimes says a node is in a cycle when it isn't). 74 * 75 * Michael Rendell, michael@stretch.cs.mun.ca - Feb 26, '90 76 */ 77 #define HASHSIZE 53 /* doesn't need to be big */ 78 #define NF_MARK 0x1 /* marker for cycle detection */ 79 #define NF_ACYCLIC 0x2 /* this node is cycle free */ 80 #define NF_NODEST 0x4 /* Unreachable */ 81 82 83 typedef struct node_str NODE; 84 85 struct node_str { 86 NODE **n_prevp; /* pointer to previous node's n_next */ 87 NODE *n_next; /* next node in graph */ 88 NODE **n_arcs; /* array of arcs to other nodes */ 89 int n_narcs; /* number of arcs in n_arcs[] */ 90 int n_arcsize; /* size of n_arcs[] array */ 91 int n_refcnt; /* # of arcs pointing to this node */ 92 int n_flags; /* NF_* */ 93 char n_name[1]; /* name of this node */ 94 }; 95 96 typedef struct _buf { 97 char *b_buf; 98 int b_bsize; 99 } BUF; 100 101 DB *db; 102 NODE *graph, **cycle_buf, **longest_cycle; 103 int debug, longest, quiet; 104 105 void add_arc __P((char *, char *)); 106 int find_cycle __P((NODE *, NODE *, int, int)); 107 NODE *get_node __P((char *)); 108 void *grow_buf __P((void *, int)); 109 void remove_node __P((NODE *)); 110 void tsort __P((void)); 111 void usage __P((void)); 112 113 int 114 main(argc, argv) 115 int argc; 116 char *argv[]; 117 { 118 register BUF *b; 119 register int c, n; 120 FILE *fp; 121 int bsize, ch, nused; 122 BUF bufs[2]; 123 124 while ((ch = getopt(argc, argv, "dlq")) != EOF) 125 switch (ch) { 126 case 'd': 127 debug = 1; 128 break; 129 case 'l': 130 longest = 1; 131 break; 132 case 'q': 133 quiet = 1; 134 break; 135 case '?': 136 default: 137 usage(); 138 } 139 argc -= optind; 140 argv += optind; 141 142 switch (argc) { 143 case 0: 144 fp = stdin; 145 break; 146 case 1: 147 if ((fp = fopen(*argv, "r")) == NULL) 148 err(1, "%s", *argv); 149 break; 150 default: 151 usage(); 152 } 153 154 for (b = bufs, n = 2; --n >= 0; b++) 155 b->b_buf = grow_buf(NULL, b->b_bsize = 1024); 156 157 /* parse input and build the graph */ 158 for (n = 0, c = getc(fp);;) { 159 while (c != EOF && isspace(c)) 160 c = getc(fp); 161 if (c == EOF) 162 break; 163 164 nused = 0; 165 b = &bufs[n]; 166 bsize = b->b_bsize; 167 do { 168 b->b_buf[nused++] = c; 169 if (nused == bsize) 170 b->b_buf = grow_buf(b->b_buf, bsize *= 2); 171 c = getc(fp); 172 } while (c != EOF && !isspace(c)); 173 174 b->b_buf[nused] = '\0'; 175 b->b_bsize = bsize; 176 if (n) 177 add_arc(bufs[0].b_buf, bufs[1].b_buf); 178 n = !n; 179 } 180 (void)fclose(fp); 181 if (n) 182 errx(1, "odd data count"); 183 184 /* do the sort */ 185 tsort(); 186 exit(0); 187 } 188 189 /* double the size of oldbuf and return a pointer to the new buffer. */ 190 void * 191 grow_buf(bp, size) 192 void *bp; 193 int size; 194 { 195 if ((bp = realloc(bp, (u_int)size)) == NULL) 196 err(1, NULL); 197 return (bp); 198 } 199 200 /* 201 * add an arc from node s1 to node s2 in the graph. If s1 or s2 are not in 202 * the graph, then add them. 203 */ 204 void 205 add_arc(s1, s2) 206 char *s1, *s2; 207 { 208 register NODE *n1; 209 NODE *n2; 210 int bsize, i; 211 212 n1 = get_node(s1); 213 214 if (!strcmp(s1, s2)) 215 return; 216 217 n2 = get_node(s2); 218 219 /* 220 * Check if this arc is already here. 221 */ 222 for (i = 0; i < n1->n_narcs; i++) 223 if (n1->n_arcs[i] == n2) 224 return; 225 /* 226 * Add it. 227 */ 228 if (n1->n_narcs == n1->n_arcsize) { 229 if (!n1->n_arcsize) 230 n1->n_arcsize = 10; 231 bsize = n1->n_arcsize * sizeof(*n1->n_arcs) * 2; 232 n1->n_arcs = grow_buf(n1->n_arcs, bsize); 233 n1->n_arcsize = bsize / sizeof(*n1->n_arcs); 234 } 235 n1->n_arcs[n1->n_narcs++] = n2; 236 ++n2->n_refcnt; 237 } 238 239 /* Find a node in the graph (insert if not found) and return a pointer to it. */ 240 NODE * 241 get_node(name) 242 char *name; 243 { 244 DBT data, key; 245 NODE *n; 246 247 if (db == NULL && 248 (db = dbopen(NULL, O_RDWR, 0, DB_HASH, NULL)) == NULL) 249 err(1, "db: %s", name); 250 251 key.data = name; 252 key.size = strlen(name) + 1; 253 254 switch ((*db->get)(db, &key, &data, 0)) { 255 case 0: 256 bcopy(data.data, &n, sizeof(n)); 257 return (n); 258 case 1: 259 break; 260 default: 261 case -1: 262 err(1, "db: %s", name); 263 } 264 265 if ((n = malloc(sizeof(NODE) + key.size)) == NULL) 266 err(1, NULL); 267 268 n->n_narcs = 0; 269 n->n_arcsize = 0; 270 n->n_arcs = NULL; 271 n->n_refcnt = 0; 272 n->n_flags = 0; 273 bcopy(name, n->n_name, key.size); 274 275 /* Add to linked list. */ 276 if ((n->n_next = graph) != NULL) 277 graph->n_prevp = &n->n_next; 278 n->n_prevp = &graph; 279 graph = n; 280 281 /* Add to hash table. */ 282 data.data = &n; 283 data.size = sizeof(n); 284 if ((*db->put)(db, &key, &data, 0)) 285 err(1, "db: %s", name); 286 return (n); 287 } 288 289 290 /* 291 * Clear the NODEST flag from all nodes. 292 */ 293 void 294 clear_cycle() 295 { 296 NODE *n; 297 298 for (n = graph; n != NULL; n = n->n_next) 299 n->n_flags &= ~NF_NODEST; 300 } 301 302 /* do topological sort on graph */ 303 void 304 tsort() 305 { 306 register NODE *n, *next; 307 register int cnt, i; 308 309 while (graph != NULL) { 310 /* 311 * Keep getting rid of simple cases until there are none left, 312 * if there are any nodes still in the graph, then there is 313 * a cycle in it. 314 */ 315 do { 316 for (cnt = 0, n = graph; n != NULL; n = next) { 317 next = n->n_next; 318 if (n->n_refcnt == 0) { 319 remove_node(n); 320 ++cnt; 321 } 322 } 323 } while (graph != NULL && cnt); 324 325 if (graph == NULL) 326 break; 327 328 if (!cycle_buf) { 329 /* 330 * Allocate space for two cycle logs - one to be used 331 * as scratch space, the other to save the longest 332 * cycle. 333 */ 334 for (cnt = 0, n = graph; n != NULL; n = n->n_next) 335 ++cnt; 336 cycle_buf = malloc((u_int)sizeof(NODE *) * cnt); 337 longest_cycle = malloc((u_int)sizeof(NODE *) * cnt); 338 if (cycle_buf == NULL || longest_cycle == NULL) 339 err(1, NULL); 340 } 341 for (n = graph; n != NULL; n = n->n_next) 342 if (!(n->n_flags & NF_ACYCLIC)) 343 if ((cnt = find_cycle(n, n, 0, 0))) { 344 if (!quiet) { 345 warnx("cycle in data"); 346 for (i = 0; i < cnt; i++) 347 warnx("%s", 348 longest_cycle[i]->n_name); 349 } 350 remove_node(n); 351 clear_cycle(); 352 break; 353 } else { 354 /* to avoid further checks */ 355 n->n_flags |= NF_ACYCLIC; 356 clear_cycle(); 357 } 358 359 if (n == NULL) 360 errx(1, "internal error -- could not find cycle"); 361 } 362 } 363 364 /* print node and remove from graph (does not actually free node) */ 365 void 366 remove_node(n) 367 register NODE *n; 368 { 369 register NODE **np; 370 register int i; 371 372 (void)printf("%s\n", n->n_name); 373 for (np = n->n_arcs, i = n->n_narcs; --i >= 0; np++) 374 --(*np)->n_refcnt; 375 n->n_narcs = 0; 376 *n->n_prevp = n->n_next; 377 if (n->n_next) 378 n->n_next->n_prevp = n->n_prevp; 379 } 380 381 382 /* look for the longest? cycle from node from to node to. */ 383 int 384 find_cycle(from, to, longest_len, depth) 385 NODE *from, *to; 386 int depth, longest_len; 387 { 388 register NODE **np; 389 register int i, len; 390 391 /* 392 * avoid infinite loops and ignore portions of the graph known 393 * to be acyclic 394 */ 395 if (from->n_flags & (NF_NODEST|NF_MARK|NF_ACYCLIC)) 396 return (0); 397 from->n_flags |= NF_MARK; 398 399 for (np = from->n_arcs, i = from->n_narcs; --i >= 0; np++) { 400 cycle_buf[depth] = *np; 401 if (*np == to) { 402 if (depth + 1 > longest_len) { 403 longest_len = depth + 1; 404 (void)memcpy((char *)longest_cycle, 405 (char *)cycle_buf, 406 longest_len * sizeof(NODE *)); 407 } 408 } else { 409 if ((*np)->n_flags & (NF_MARK|NF_ACYCLIC|NF_NODEST)) 410 continue; 411 len = find_cycle(*np, to, longest_len, depth + 1); 412 413 if (debug) 414 (void)printf("%*s %s->%s %d\n", depth, "", 415 from->n_name, to->n_name, len); 416 417 if (len == 0) 418 (*np)->n_flags |= NF_NODEST; 419 420 if (len > longest_len) 421 longest_len = len; 422 423 if (len > 0 && !longest) 424 break; 425 } 426 } 427 from->n_flags &= ~NF_MARK; 428 return (longest_len); 429 } 430 431 void 432 usage() 433 { 434 (void)fprintf(stderr, "usage: tsort [-dlq] [file]\n"); 435 exit(1); 436 } 437