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