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 * $FreeBSD$ 37 */ 38 39 #ifndef lint 40 static const 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 const 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 #include <unistd.h> 60 61 /* 62 * Topological sort. Input is a list of pairs of strings separated by 63 * white space (spaces, tabs, and/or newlines); strings are written to 64 * standard output in sorted order, one per line. 65 * 66 * usage: 67 * tsort [-dlq] [inputfile] 68 * If no input file is specified, standard input is read. 69 * 70 * Should be compatible with AT&T tsort HOWEVER the output is not identical 71 * (i.e. for most graphs there is more than one sorted order, and this tsort 72 * usually generates a different one then the AT&T tsort). Also, cycle 73 * reporting seems to be more accurate in this version (the AT&T tsort 74 * sometimes says a node is in a cycle when it isn't). 75 * 76 * Michael Rendell, michael@stretch.cs.mun.ca - Feb 26, '90 77 */ 78 79 #define NF_MARK 0x1 /* marker for cycle detection */ 80 #define NF_ACYCLIC 0x2 /* this node is cycle free */ 81 #define NF_NODEST 0x4 /* Unreachable */ 82 83 84 typedef struct node_str NODE; 85 86 struct node_str { 87 NODE **n_prevp; /* pointer to previous node's n_next */ 88 NODE *n_next; /* next node in graph */ 89 NODE **n_arcs; /* array of arcs to other nodes */ 90 int n_narcs; /* number of arcs in n_arcs[] */ 91 int n_arcsize; /* size of n_arcs[] array */ 92 int n_refcnt; /* # of arcs pointing to this node */ 93 int n_flags; /* NF_* */ 94 char n_name[1]; /* name of this node */ 95 }; 96 97 typedef struct _buf { 98 char *b_buf; 99 int b_bsize; 100 } BUF; 101 102 DB *db; 103 NODE *graph, **cycle_buf, **longest_cycle; 104 int debug, longest, quiet; 105 106 void add_arc(char *, char *); 107 int find_cycle(NODE *, NODE *, int, int); 108 NODE *get_node(char *); 109 void *grow_buf(void *, size_t); 110 void remove_node(NODE *); 111 void clear_cycle(void); 112 void tsort(void); 113 void usage(void); 114 115 int 116 main(int argc, char *argv[]) 117 { 118 BUF *b; 119 int c, n; 120 FILE *fp; 121 int bsize, ch, nused; 122 BUF bufs[2]; 123 124 while ((ch = getopt(argc, argv, "dlq")) != -1) 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(void *bp, size_t size) 192 { 193 if ((bp = realloc(bp, size)) == NULL) 194 err(1, NULL); 195 return (bp); 196 } 197 198 /* 199 * add an arc from node s1 to node s2 in the graph. If s1 or s2 are not in 200 * the graph, then add them. 201 */ 202 void 203 add_arc(char *s1, char *s2) 204 { 205 NODE *n1; 206 NODE *n2; 207 int bsize, i; 208 209 n1 = get_node(s1); 210 211 if (!strcmp(s1, s2)) 212 return; 213 214 n2 = get_node(s2); 215 216 /* 217 * Check if this arc is already here. 218 */ 219 for (i = 0; i < n1->n_narcs; i++) 220 if (n1->n_arcs[i] == n2) 221 return; 222 /* 223 * Add it. 224 */ 225 if (n1->n_narcs == n1->n_arcsize) { 226 if (!n1->n_arcsize) 227 n1->n_arcsize = 10; 228 bsize = n1->n_arcsize * sizeof(*n1->n_arcs) * 2; 229 n1->n_arcs = grow_buf(n1->n_arcs, bsize); 230 n1->n_arcsize = bsize / sizeof(*n1->n_arcs); 231 } 232 n1->n_arcs[n1->n_narcs++] = n2; 233 ++n2->n_refcnt; 234 } 235 236 /* Find a node in the graph (insert if not found) and return a pointer to it. */ 237 NODE * 238 get_node(char *name) 239 { 240 DBT data, key; 241 NODE *n; 242 243 if (db == NULL && 244 (db = dbopen(NULL, O_RDWR, 0, DB_HASH, NULL)) == NULL) 245 err(1, "db: %s", name); 246 247 key.data = name; 248 key.size = strlen(name) + 1; 249 250 switch ((*db->get)(db, &key, &data, 0)) { 251 case 0: 252 bcopy(data.data, &n, sizeof(n)); 253 return (n); 254 case 1: 255 break; 256 default: 257 case -1: 258 err(1, "db: %s", name); 259 } 260 261 if ((n = malloc(sizeof(NODE) + key.size)) == NULL) 262 err(1, NULL); 263 264 n->n_narcs = 0; 265 n->n_arcsize = 0; 266 n->n_arcs = NULL; 267 n->n_refcnt = 0; 268 n->n_flags = 0; 269 bcopy(name, n->n_name, key.size); 270 271 /* Add to linked list. */ 272 if ((n->n_next = graph) != NULL) 273 graph->n_prevp = &n->n_next; 274 n->n_prevp = &graph; 275 graph = n; 276 277 /* Add to hash table. */ 278 data.data = &n; 279 data.size = sizeof(n); 280 if ((*db->put)(db, &key, &data, 0)) 281 err(1, "db: %s", name); 282 return (n); 283 } 284 285 286 /* 287 * Clear the NODEST flag from all nodes. 288 */ 289 void 290 clear_cycle(void) 291 { 292 NODE *n; 293 294 for (n = graph; n != NULL; n = n->n_next) 295 n->n_flags &= ~NF_NODEST; 296 } 297 298 /* do topological sort on graph */ 299 void 300 tsort(void) 301 { 302 NODE *n, *next; 303 int cnt, i; 304 305 while (graph != NULL) { 306 /* 307 * Keep getting rid of simple cases until there are none left, 308 * if there are any nodes still in the graph, then there is 309 * a cycle in it. 310 */ 311 do { 312 for (cnt = 0, n = graph; n != NULL; n = next) { 313 next = n->n_next; 314 if (n->n_refcnt == 0) { 315 remove_node(n); 316 ++cnt; 317 } 318 } 319 } while (graph != NULL && cnt); 320 321 if (graph == NULL) 322 break; 323 324 if (!cycle_buf) { 325 /* 326 * Allocate space for two cycle logs - one to be used 327 * as scratch space, the other to save the longest 328 * cycle. 329 */ 330 for (cnt = 0, n = graph; n != NULL; n = n->n_next) 331 ++cnt; 332 cycle_buf = malloc(sizeof(NODE *) * cnt); 333 longest_cycle = malloc(sizeof(NODE *) * cnt); 334 if (cycle_buf == NULL || longest_cycle == NULL) 335 err(1, NULL); 336 } 337 for (n = graph; n != NULL; n = n->n_next) 338 if (!(n->n_flags & NF_ACYCLIC)) { 339 if ((cnt = find_cycle(n, n, 0, 0))) { 340 if (!quiet) { 341 warnx("cycle in data"); 342 for (i = 0; i < cnt; i++) 343 warnx("%s", 344 longest_cycle[i]->n_name); 345 } 346 remove_node(n); 347 clear_cycle(); 348 break; 349 } else { 350 /* to avoid further checks */ 351 n->n_flags |= NF_ACYCLIC; 352 clear_cycle(); 353 } 354 } 355 356 if (n == NULL) 357 errx(1, "internal error -- could not find cycle"); 358 } 359 } 360 361 /* print node and remove from graph (does not actually free node) */ 362 void 363 remove_node(NODE *n) 364 { 365 NODE **np; 366 int i; 367 368 (void)printf("%s\n", n->n_name); 369 for (np = n->n_arcs, i = n->n_narcs; --i >= 0; np++) 370 --(*np)->n_refcnt; 371 n->n_narcs = 0; 372 *n->n_prevp = n->n_next; 373 if (n->n_next) 374 n->n_next->n_prevp = n->n_prevp; 375 } 376 377 378 /* look for the longest? cycle from node from to node to. */ 379 int 380 find_cycle(NODE *from, NODE *to, int longest_len, int depth) 381 { 382 NODE **np; 383 int i, len; 384 385 /* 386 * avoid infinite loops and ignore portions of the graph known 387 * to be acyclic 388 */ 389 if (from->n_flags & (NF_NODEST|NF_MARK|NF_ACYCLIC)) 390 return (0); 391 from->n_flags |= NF_MARK; 392 393 for (np = from->n_arcs, i = from->n_narcs; --i >= 0; np++) { 394 cycle_buf[depth] = *np; 395 if (*np == to) { 396 if (depth + 1 > longest_len) { 397 longest_len = depth + 1; 398 (void)memcpy((char *)longest_cycle, 399 (char *)cycle_buf, 400 longest_len * sizeof(NODE *)); 401 } 402 } else { 403 if ((*np)->n_flags & (NF_MARK|NF_ACYCLIC|NF_NODEST)) 404 continue; 405 len = find_cycle(*np, to, longest_len, depth + 1); 406 407 if (debug) 408 (void)printf("%*s %s->%s %d\n", depth, "", 409 from->n_name, to->n_name, len); 410 411 if (len == 0) 412 (*np)->n_flags |= NF_NODEST; 413 414 if (len > longest_len) 415 longest_len = len; 416 417 if (len > 0 && !longest) 418 break; 419 } 420 } 421 from->n_flags &= ~NF_MARK; 422 return (longest_len); 423 } 424 425 void 426 usage(void) 427 { 428 (void)fprintf(stderr, "usage: tsort [-dlq] [file]\n"); 429 exit(1); 430 } 431