2 * Copyright (c) 1989, 1993, 1994
3 * The Regents of the University of California. All rights reserved.
5 * This code is derived from software contributed to Berkeley by
6 * Michael Rendell of Memorial University of Newfoundland.
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
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 * 4. 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.
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
34 static const char copyright[] =
35 "@(#) Copyright (c) 1989, 1993, 1994\n\
36 The Regents of the University of California. All rights reserved.\n";
40 static const char sccsid[] = "@(#)tsort.c 8.3 (Berkeley) 5/4/95";
43 #include <sys/cdefs.h>
44 __FBSDID("$FreeBSD$");
46 #include <sys/types.h>
59 * Topological sort. Input is a list of pairs of strings separated by
60 * white space (spaces, tabs, and/or newlines); strings are written to
61 * standard output in sorted order, one per line.
64 * tsort [-dlq] [inputfile]
65 * If no input file is specified, standard input is read.
67 * Should be compatible with AT&T tsort HOWEVER the output is not identical
68 * (i.e. for most graphs there is more than one sorted order, and this tsort
69 * usually generates a different one then the AT&T tsort). Also, cycle
70 * reporting seems to be more accurate in this version (the AT&T tsort
71 * sometimes says a node is in a cycle when it isn't).
73 * Michael Rendell, michael@stretch.cs.mun.ca - Feb 26, '90
76 #define NF_MARK 0x1 /* marker for cycle detection */
77 #define NF_ACYCLIC 0x2 /* this node is cycle free */
78 #define NF_NODEST 0x4 /* Unreachable */
81 typedef struct node_str NODE;
84 NODE **n_prevp; /* pointer to previous node's n_next */
85 NODE *n_next; /* next node in graph */
86 NODE **n_arcs; /* array of arcs to other nodes */
87 int n_narcs; /* number of arcs in n_arcs[] */
88 int n_arcsize; /* size of n_arcs[] array */
89 int n_refcnt; /* # of arcs pointing to this node */
90 int n_flags; /* NF_* */
91 char n_name[1]; /* name of this node */
100 static NODE *graph, **cycle_buf, **longest_cycle;
101 static int debug, longest, quiet;
103 static void add_arc(char *, char *);
104 static int find_cycle(NODE *, NODE *, int, int);
105 static NODE *get_node(char *);
106 static void *grow_buf(void *, size_t);
107 static void remove_node(NODE *);
108 static void clear_cycle(void);
109 static void tsort(void);
110 static void usage(void);
113 main(int argc, char *argv[])
118 int bsize, ch, nused;
122 while ((ch = getopt(argc, argv, "dlq")) != -1)
145 if ((fp = fopen(*argv, "r")) == NULL)
152 for (b = bufs, n = 2; --n >= 0; b++)
153 b->b_buf = grow_buf(NULL, b->b_bsize = 1024);
155 /* parse input and build the graph */
156 for (n = 0, c = getc(fp);;) {
157 while (c != EOF && isspace(c))
166 b->b_buf[nused++] = c;
168 b->b_buf = grow_buf(b->b_buf, bsize *= 2);
170 } while (c != EOF && !isspace(c));
172 b->b_buf[nused] = '\0';
175 add_arc(bufs[0].b_buf, bufs[1].b_buf);
180 errx(1, "odd data count");
187 /* double the size of oldbuf and return a pointer to the new buffer. */
189 grow_buf(void *bp, size_t size)
191 if ((bp = realloc(bp, size)) == NULL)
197 * add an arc from node s1 to node s2 in the graph. If s1 or s2 are not in
198 * the graph, then add them.
201 add_arc(char *s1, char *s2)
215 * Check if this arc is already here.
217 for (i = 0; i < n1->n_narcs; i++)
218 if (n1->n_arcs[i] == n2)
223 if (n1->n_narcs == n1->n_arcsize) {
226 bsize = n1->n_arcsize * sizeof(*n1->n_arcs) * 2;
227 n1->n_arcs = grow_buf(n1->n_arcs, bsize);
228 n1->n_arcsize = bsize / sizeof(*n1->n_arcs);
230 n1->n_arcs[n1->n_narcs++] = n2;
234 /* Find a node in the graph (insert if not found) and return a pointer to it. */
242 (db = dbopen(NULL, O_RDWR, 0, DB_HASH, NULL)) == NULL)
243 err(1, "db: %s", name);
246 key.size = strlen(name) + 1;
248 switch ((*db->get)(db, &key, &data, 0)) {
250 bcopy(data.data, &n, sizeof(n));
256 err(1, "db: %s", name);
259 if ((n = malloc(sizeof(NODE) + key.size)) == NULL)
267 bcopy(name, n->n_name, key.size);
269 /* Add to linked list. */
270 if ((n->n_next = graph) != NULL)
271 graph->n_prevp = &n->n_next;
275 /* Add to hash table. */
277 data.size = sizeof(n);
278 if ((*db->put)(db, &key, &data, 0))
279 err(1, "db: %s", name);
285 * Clear the NODEST flag from all nodes.
292 for (n = graph; n != NULL; n = n->n_next)
293 n->n_flags &= ~NF_NODEST;
296 /* do topological sort on graph */
303 while (graph != NULL) {
305 * Keep getting rid of simple cases until there are none left,
306 * if there are any nodes still in the graph, then there is
310 for (cnt = 0, n = graph; n != NULL; n = next) {
312 if (n->n_refcnt == 0) {
317 } while (graph != NULL && cnt);
324 * Allocate space for two cycle logs - one to be used
325 * as scratch space, the other to save the longest
328 for (cnt = 0, n = graph; n != NULL; n = n->n_next)
330 cycle_buf = malloc(sizeof(NODE *) * cnt);
331 longest_cycle = malloc(sizeof(NODE *) * cnt);
332 if (cycle_buf == NULL || longest_cycle == NULL)
335 for (n = graph; n != NULL; n = n->n_next)
336 if (!(n->n_flags & NF_ACYCLIC)) {
337 if ((cnt = find_cycle(n, n, 0, 0))) {
339 warnx("cycle in data");
340 for (i = 0; i < cnt; i++)
342 longest_cycle[i]->n_name);
348 /* to avoid further checks */
349 n->n_flags |= NF_ACYCLIC;
355 errx(1, "internal error -- could not find cycle");
359 /* print node and remove from graph (does not actually free node) */
366 (void)printf("%s\n", n->n_name);
367 for (np = n->n_arcs, i = n->n_narcs; --i >= 0; np++)
370 *n->n_prevp = n->n_next;
372 n->n_next->n_prevp = n->n_prevp;
376 /* look for the longest? cycle from node from to node to. */
378 find_cycle(NODE *from, NODE *to, int longest_len, int depth)
384 * avoid infinite loops and ignore portions of the graph known
387 if (from->n_flags & (NF_NODEST|NF_MARK|NF_ACYCLIC))
389 from->n_flags |= NF_MARK;
391 for (np = from->n_arcs, i = from->n_narcs; --i >= 0; np++) {
392 cycle_buf[depth] = *np;
394 if (depth + 1 > longest_len) {
395 longest_len = depth + 1;
396 (void)memcpy((char *)longest_cycle,
398 longest_len * sizeof(NODE *));
401 if ((*np)->n_flags & (NF_MARK|NF_ACYCLIC|NF_NODEST))
403 len = find_cycle(*np, to, longest_len, depth + 1);
406 (void)printf("%*s %s->%s %d\n", depth, "",
407 from->n_name, to->n_name, len);
410 (*np)->n_flags |= NF_NODEST;
412 if (len > longest_len)
415 if (len > 0 && !longest)
419 from->n_flags &= ~NF_MARK;
420 return (longest_len);
426 (void)fprintf(stderr, "usage: tsort [-dlq] [file]\n");