|
|
1.1 root 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 are permitted provided
9: * that: (1) source distributions retain this entire copyright notice and
10: * comment, and (2) distributions including binaries display the following
11: * acknowledgement: ``This product includes software developed by the
12: * University of California, Berkeley and its contributors'' in the
13: * documentation or other materials provided with the distribution and in
14: * all advertising materials mentioning features or use of this software.
15: * Neither the name of the University nor the names of its contributors may
16: * be used to endorse or promote products derived from this software without
17: * specific prior written permission.
18: * THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR IMPLIED
19: * WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF
20: * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.
21: */
22:
23: #ifndef lint
24: char copyright[] =
25: "@(#) Copyright (c) 1989 The Regents of the University of California.\n\
26: All rights reserved.\n";
27: #endif /* not lint */
28:
29: #ifndef lint
30: static char sccsid[] = "@(#)tsort.c 5.3 (Berkeley) 6/1/90";
31: #endif /* not lint */
32:
33: #include <sys/types.h>
34: #include <errno.h>
35: #include <stdio.h>
36: #include <ctype.h>
37: #include <string.h>
38:
39: /*
40: * Topological sort. Input is a list of pairs of strings seperated by
41: * white space (spaces, tabs, and/or newlines); strings are written to
42: * standard output in sorted order, one per line.
43: *
44: * usage:
45: * tsort [inputfile]
46: * If no input file is specified, standard input is read.
47: *
48: * Should be compatable with AT&T tsort HOWEVER the output is not identical
49: * (i.e. for most graphs there is more than one sorted order, and this tsort
50: * usually generates a different one then the AT&T tsort). Also, cycle
51: * reporting seems to be more accurate in this version (the AT&T tsort
52: * sometimes says a node is in a cycle when it isn't).
53: *
54: * Michael Rendell, [email protected] - Feb 26, '90
55: */
56: #define HASHSIZE 53 /* doesn't need to be big */
57: #define NF_MARK 0x1 /* marker for cycle detection */
58: #define NF_ACYCLIC 0x2 /* this node is cycle free */
59:
60: typedef struct node_str NODE;
61:
62: struct node_str {
63: char *n_name; /* name of this node */
64: NODE **n_prevp; /* pointer to previous node's n_next */
65: NODE *n_next; /* next node in graph */
66: NODE *n_hash; /* next node in hash table */
67: int n_narcs; /* number of arcs in n_arcs[] */
68: int n_arcsize; /* size of n_arcs[] array */
69: NODE **n_arcs; /* array of arcs to other nodes */
70: int n_refcnt; /* # of arcs pointing to this node */
71: int n_flags; /* NF_* */
72: };
73:
74: typedef struct _buf {
75: char *b_buf;
76: int b_bsize;
77: } BUF;
78:
79: NODE *add_node(), *find_node();
80: void add_arc(), no_memory(), remove_node(), tsort();
81: char *grow_buf(), *malloc();
82:
83: extern int errno;
84: NODE *graph;
85: NODE *hashtable[HASHSIZE];
86: NODE **cycle_buf;
87: NODE **longest_cycle;
88:
89: main(argc, argv)
90: int argc;
91: char **argv;
92: {
93: register BUF *b;
94: register int c, n;
95: FILE *fp;
96: int bsize, nused;
97: BUF bufs[2];
98:
99: if (argc < 2)
100: fp = stdin;
101: else if (argc == 2) {
102: (void)fprintf(stderr, "usage: tsort [ inputfile ]\n");
103: exit(1);
104: } else if (!(fp = fopen(argv[1], "r"))) {
105: (void)fprintf(stderr, "tsort: %s.\n", strerror(errno));
106: exit(1);
107: }
108:
109: for (b = bufs, n = 2; --n >= 0; b++)
110: b->b_buf = grow_buf((char *)NULL, b->b_bsize = 1024);
111:
112: /* parse input and build the graph */
113: for (n = 0, c = getc(fp);;) {
114: while (c != EOF && isspace(c))
115: c = getc(fp);
116: if (c == EOF)
117: break;
118:
119: nused = 0;
120: b = &bufs[n];
121: bsize = b->b_bsize;
122: do {
123: b->b_buf[nused++] = c;
124: if (nused == bsize) {
125: bsize *= 2;
126: b->b_buf = grow_buf(b->b_buf, bsize);
127: }
128: c = getc(fp);
129: } while (c != EOF && !isspace(c));
130:
131: b->b_buf[nused] = '\0';
132: b->b_bsize = bsize;
133: if (n)
134: add_arc(bufs[0].b_buf, bufs[1].b_buf);
135: n = !n;
136: }
137: (void)fclose(fp);
138: if (n) {
139: (void)fprintf(stderr, "tsort: odd data count.\n");
140: exit(1);
141: }
142:
143: /* do the sort */
144: tsort();
145: exit(0);
146: }
147:
148: /* double the size of oldbuf and return a pointer to the new buffer. */
149: char *
150: grow_buf(bp, size)
151: char *bp;
152: int size;
153: {
154: char *realloc();
155:
156: if (!(bp = realloc(bp, (u_int)size)))
157: no_memory();
158: return(bp);
159: }
160:
161: /*
162: * add an arc from node s1 to node s2 in the graph. If s1 or s2 are not in
163: * the graph, then add them.
164: */
165: void
166: add_arc(s1, s2)
167: char *s1, *s2;
168: {
169: register NODE *n1;
170: NODE *n2;
171: int bsize;
172:
173: n1 = find_node(s1);
174: if (!n1)
175: n1 = add_node(s1);
176:
177: if (!strcmp(s1, s2))
178: return;
179:
180: n2 = find_node(s2);
181: if (!n2)
182: n2 = add_node(s2);
183:
184: /*
185: * could check to see if this arc is here already, but it isn't
186: * worth the bother -- there usually isn't and it doesn't hurt if
187: * there is (I think :-).
188: */
189: if (n1->n_narcs == n1->n_arcsize) {
190: if (!n1->n_arcsize)
191: n1->n_arcsize = 10;
192: bsize = n1->n_arcsize * sizeof(*n1->n_arcs) * 2;
193: n1->n_arcs = (NODE **)grow_buf((char *)n1->n_arcs, bsize);
194: n1->n_arcsize = bsize / sizeof(*n1->n_arcs);
195: }
196: n1->n_arcs[n1->n_narcs++] = n2;
197: ++n2->n_refcnt;
198: }
199:
200: hash_string(s)
201: char *s;
202: {
203: register int hash, i;
204:
205: for (hash = 0, i = 1; *s; s++, i++)
206: hash += *s * i;
207: return(hash % HASHSIZE);
208: }
209:
210: /*
211: * find a node in the graph and return a pointer to it - returns null if not
212: * found.
213: */
214: NODE *
215: find_node(name)
216: char *name;
217: {
218: register NODE *n;
219:
220: for (n = hashtable[hash_string(name)]; n; n = n->n_hash)
221: if (!strcmp(n->n_name, name))
222: return(n);
223: return((NODE *)NULL);
224: }
225:
226: /* Add a node to the graph and return a pointer to it. */
227: NODE *
228: add_node(name)
229: char *name;
230: {
231: register NODE *n;
232: int hash;
233:
234: if (!(n = (NODE *)malloc(sizeof(NODE))) || !(n->n_name = strdup(name)))
235: no_memory();
236:
237: n->n_narcs = 0;
238: n->n_arcsize = 0;
239: n->n_arcs = (NODE **)NULL;
240: n->n_refcnt = 0;
241: n->n_flags = 0;
242:
243: /* add to linked list */
244: if (n->n_next = graph)
245: graph->n_prevp = &n->n_next;
246: n->n_prevp = &graph;
247: graph = n;
248:
249: /* add to hash table */
250: hash = hash_string(name);
251: n->n_hash = hashtable[hash];
252: hashtable[hash] = n;
253: return(n);
254: }
255:
256: /* do topological sort on graph */
257: void
258: tsort()
259: {
260: register NODE *n, *next;
261: register int cnt;
262:
263: while (graph) {
264: /*
265: * keep getting rid of simple cases until there are none left,
266: * if there are any nodes still in the graph, then there is
267: * a cycle in it.
268: */
269: do {
270: for (cnt = 0, n = graph; n; n = next) {
271: next = n->n_next;
272: if (n->n_refcnt == 0) {
273: remove_node(n);
274: ++cnt;
275: }
276: }
277: } while (graph && cnt);
278:
279: if (!graph)
280: break;
281:
282: if (!cycle_buf) {
283: /*
284: * allocate space for two cycle logs - one to be used
285: * as scratch space, the other to save the longest
286: * cycle.
287: */
288: for (cnt = 0, n = graph; n; n = n->n_next)
289: ++cnt;
290: cycle_buf =
291: (NODE **)malloc((u_int)sizeof(NODE *) * cnt);
292: longest_cycle =
293: (NODE **)malloc((u_int)sizeof(NODE *) * cnt);
294: if (!cycle_buf || !longest_cycle)
295: no_memory();
296: }
297: for (n = graph; n; n = n->n_next)
298: if (!(n->n_flags & NF_ACYCLIC)) {
299: if (cnt = find_cycle(n, n, 0, 0)) {
300: register int i;
301:
302: (void)fprintf(stderr,
303: "tsort: cycle in data.\n");
304: for (i = 0; i < cnt; i++)
305: (void)fprintf(stderr,
306: "tsort: %s.\n", longest_cycle[i]->n_name);
307: remove_node(n);
308: break;
309: } else
310: /* to avoid further checks */
311: n->n_flags = NF_ACYCLIC;
312: }
313:
314: if (!n) {
315: (void)fprintf(stderr,
316: "tsort: internal error -- could not find cycle.\n");
317: exit(1);
318: }
319: }
320: }
321:
322: /* print node and remove from graph (does not actually free node) */
323: void
324: remove_node(n)
325: register NODE *n;
326: {
327: register NODE **np;
328: register int i;
329:
330: (void)printf("%s\n", n->n_name);
331: for (np = n->n_arcs, i = n->n_narcs; --i >= 0; np++)
332: --(*np)->n_refcnt;
333: n->n_narcs = 0;
334: *n->n_prevp = n->n_next;
335: if (n->n_next)
336: n->n_next->n_prevp = n->n_prevp;
337: }
338:
339: /* look for the longest cycle from node from to node to. */
340: find_cycle(from, to, longest_len, depth)
341: NODE *from, *to;
342: int depth, longest_len;
343: {
344: register NODE **np;
345: register int i, len;
346:
347: /*
348: * avoid infinite loops and ignore portions of the graph known
349: * to be acyclic
350: */
351: if (from->n_flags & (NF_MARK|NF_ACYCLIC))
352: return(0);
353: from->n_flags = NF_MARK;
354:
355: for (np = from->n_arcs, i = from->n_narcs; --i >= 0; np++) {
356: cycle_buf[depth] = *np;
357: if (*np == to) {
358: if (depth + 1 > longest_len) {
359: longest_len = depth + 1;
360: (void)memcpy((char *)longest_cycle,
361: (char *)cycle_buf,
362: longest_len * sizeof(NODE *));
363: }
364: } else {
365: len = find_cycle(*np, to, longest_len, depth + 1);
366: if (len > longest_len)
367: longest_len = len;
368: }
369: }
370: from->n_flags &= ~NF_MARK;
371: return(longest_len);
372: }
373:
374: void
375: no_memory()
376: {
377: (void)fprintf(stderr, "tsort: %s.\n", strerror(ENOMEM));
378: exit(1);
379: }
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.