xref: /illumos-gate/usr/src/cmd/diff3/diff3prog.c (revision a6d42e7d71324c5193c3b94d57d96ba2925d52e1)
1 /*
2  * CDDL HEADER START
3  *
4  * The contents of this file are subject to the terms of the
5  * Common Development and Distribution License (the "License").
6  * You may not use this file except in compliance with the License.
7  *
8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9  * or http://www.opensolaris.org/os/licensing.
10  * See the License for the specific language governing permissions
11  * and limitations under the License.
12  *
13  * When distributing Covered Code, include this CDDL HEADER in each
14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15  * If applicable, add the following below this CDDL HEADER, with the
16  * fields enclosed by brackets "[]" replaced with your own identifying
17  * information: Portions Copyright [yyyy] [name of copyright owner]
18  *
19  * CDDL HEADER END
20  */
21 /*	Copyright (c) 1984, 1986, 1987, 1988, 1989 AT&T	*/
22 /*	  All Rights Reserved  	*/
23 
24 
25 /*
26  * Copyright 2006 Sun Microsystems, Inc.  All rights reserved.
27  * Use is subject to license terms.
28  */
29 
30 #pragma ident	"%Z%%M%	%I%	%E% SMI"
31 
32 #include <stdio.h>
33 #include <stdlib.h>
34 #include <unistd.h>
35 #include <limits.h>
36 #include <sys/param.h>
37 #include <errno.h>
38 #
39 
40 /*
41  * diff3 - 3-way differential file comparison
42  */
43 
44 /*
45  * diff3 [-ex3EX] d13 d23 f1 f2 f3 [m1 m3]
46  *
47  * d13 = diff report on f1 vs f3
48  * d23 = diff report on f2 vs f3
49  * f1, f2, f3 the 3 files
50  * if changes in f1 overlap with changes in f3, m1 and m3 are used
51  * to mark the overlaps; otherwise, the file names f1 and f3 are used
52  * (only for options E and X).
53  */
54 
55 struct  range {int from, to; };
56 	/*
57 	 * from is first in range of changed lines
58 	 * to is last+1
59 	 * from = to = line after point of insertion
60 	 * for added lines
61 	 */
62 struct diff {struct range old, new; };
63 
64 #define	NC 4096
65 /*
66  * de is used to gather editing scripts,
67  * that are later spewed out in reverse order.
68  * its first element must be all zero
69  * the "new" component of de contains line positions
70  * or byte positions depending on when you look(!?)
71  */
72 static struct diff d13[NC];
73 static struct diff d23[NC];
74 
75 /*
76  * array overlap indicates which sections in de correspond to
77  * lines that are different in all three files.
78  */
79 
80 static struct diff de[NC];
81 static char overlap[NC];
82 static int  overlapcnt = 0;
83 
84 static char line[LINE_MAX+1];
85 static FILE *fp[3];
86 /*
87  *	the number of the last-read line in each file
88  *	is kept in cline[0-2]
89  */
90 static int cline[3];
91 /*
92  *	the latest known correspondence between line
93  *	numbers of the 3 files is stored in last[1-3]
94  */
95 static int last[4];
96 static int eflag;
97 static int oflag;	/* indicates whether to mark overlaps (-E or -X) */
98 static int debug  = 0;
99 /* markers for -E and -X: */
100 static char f1mark[8+MAXPATHLEN], f3mark[8+MAXPATHLEN];
101 		/* Need space for "<<<<<<< " or ">>>>>>> " plus filename   */
102 static int save_err;	/* saves errno */
103 
104 static int readin(char *name, struct diff *dd);
105 static int number(char **lc);
106 static int digit(int c);
107 static int getchange(FILE *b);
108 static int getline(FILE *b);
109 static void merge(int m1, int m2);
110 static void separate(char *s);
111 static void change(int i, struct range *rold, int dup);
112 static void prange(struct range *rold);
113 static void keep(int i, struct range *rnew);
114 static int skip(int i, int from, char *pr);
115 static int duplicate(struct range *r1, struct range *r2);
116 static void repos(int nchar);
117 static void trouble();
118 static int edit(struct diff *diff, int dup, int j);
119 static void edscript(int n);
120 static void usage();
121 
122 int
123 main(int argc, char **argv)
124 {
125 	int i, m, n;
126 	eflag  = 0;
127 	oflag  = 0;
128 	if ((argc > 1) && (*argv[1] == '-')) {
129 		switch (argv[1][1]) {
130 		case 'e':
131 			eflag = 3;
132 			break;
133 		case '3':
134 			eflag = 2;
135 			break;
136 		case 'x':
137 			eflag = 1;
138 			break;
139 		case 'E':
140 			eflag = 3;
141 			oflag = 1;
142 			break;
143 		case 'X':
144 			oflag = eflag = 1;
145 			break;
146 		default:
147 			usage();
148 			break;
149 		}
150 		argv++;
151 		argc--;
152 	}
153 	if (argc < 6)
154 		usage();
155 	if (oflag) {
156 		(void) snprintf(f1mark, sizeof (f1mark), "<<<<<<< %s",
157 						argc >= 7 ? argv[6] : argv[3]);
158 		(void) snprintf(f3mark, sizeof (f3mark), ">>>>>>> %s",
159 						argc >= 8 ? argv[7] : argv[5]);
160 	}
161 
162 	m = readin(argv[1], d13);
163 	n = readin(argv[2], d23);
164 	for (i = 0; i <= 2; i++)
165 		if ((fp[i] = fopen(argv[i+3], "r")) == NULL) {
166 			save_err = errno;
167 			(void) fprintf(stderr, "diff3: can't open %s: ",
168 				argv[i+3]);
169 			errno = save_err;
170 			perror("");
171 			exit(1);
172 		}
173 	merge(m, n);
174 	return (0);
175 }
176 
177 /*
178  * pick up the line numbers of all changes from
179  * one change file
180  * (this puts the numbers in a vector, which is not
181  * strictly necessary, since the vector is processed
182  * in one sequential pass. The vector could be optimized
183  * out of existence)
184  */
185 
186 static int
187 readin(char *name, struct diff *dd)
188 {
189 	int i;
190 	int a, b, c, d;
191 	char kind;
192 	char *p;
193 	if ((fp[0] = fopen(name, "r")) == NULL) {
194 		save_err = errno;
195 		(void) fprintf(stderr, "diff3: can't open %s: ", name);
196 		errno = save_err;
197 		perror("");
198 		exit(1);
199 	}
200 	for (i = 0; getchange(fp[0]); i++) {
201 		if (i >= NC) {
202 			(void) fprintf(stderr, "diff3: too many changes\n");
203 			exit(0);
204 		}
205 		p = line;
206 		a = b = number(&p);
207 		if (*p == ',') {
208 			p++;
209 			b = number(&p);
210 		}
211 		kind = *p++;
212 		c = d = number(&p);
213 		if (*p == ',') {
214 			p++;
215 			d = number(&p);
216 		}
217 		if (kind == 'a')
218 			a++;
219 		if (kind == 'd')
220 			c++;
221 		b++;
222 		d++;
223 		dd[i].old.from = a;
224 		dd[i].old.to = b;
225 		dd[i].new.from = c;
226 		dd[i].new.to = d;
227 	}
228 	dd[i].old.from = dd[i-1].old.to;
229 	dd[i].new.from = dd[i-1].new.to;
230 	(void) fclose(fp[0]);
231 	return (i);
232 }
233 
234 static int
235 number(char **lc)
236 {
237 	int nn;
238 	nn = 0;
239 	while (digit(**lc))
240 		nn = nn*10 + *(*lc)++ - '0';
241 	return (nn);
242 }
243 
244 static int
245 digit(int c)
246 {
247 	return (c >= '0' && c <= '9');
248 }
249 
250 static int
251 getchange(FILE *b)
252 {
253 	while (getline(b))
254 		if (digit(line[0]))
255 			return (1);
256 	return (0);
257 }
258 
259 static int
260 getline(FILE *b)
261 {
262 	int i, c;
263 	for (i = 0; i < sizeof (line)-1; i++) {
264 		c = getc(b);
265 		if (c == EOF) {
266 			line[i] = 0;
267 			return (i);
268 		}
269 		line[i] = c;
270 		if (c == '\n') {
271 			line[++i] = 0;
272 			return (i);
273 		}
274 	}
275 	return (0);
276 }
277 
278 static void
279 merge(int m1, int m2)
280 {
281 	struct diff *d1, *d2, *d3;
282 	int dup;
283 	int j;
284 	int t1, t2;
285 	d1 = d13;
286 	d2 = d23;
287 	j = 0;
288 	for (; (t1 = d1 < d13+m1) | (t2 = d2 < d23+m2); ) {
289 		if (debug) {
290 			(void) printf("%d,%d=%d,%d %d,%d=%d,%d\n",
291 			d1->old.from, d1->old.to,
292 			d1->new.from, d1->new.to,
293 			d2->old.from, d2->old.to,
294 			d2->new.from, d2->new.to);
295 		}
296 
297 		/* first file is different from others */
298 		if (!t2 || t1 && d1->new.to < d2->new.from) {
299 			/* stuff peculiar to 1st file */
300 			if (eflag == 0) {
301 				separate("1");
302 				change(1, &d1->old, 0);
303 				keep(2, &d1->new);
304 				change(3, &d1->new, 0);
305 			}
306 			d1++;
307 			continue;
308 		}
309 
310 		/* second file is different from others */
311 		if (!t1 || t2 && d2->new.to < d1->new.from) {
312 			if (eflag == 0) {
313 				separate("2");
314 				keep(1, &d2->new);
315 				change(2, &d2->old, 0);
316 				change(3, &d2->new, 0);
317 			}
318 			d2++;
319 			continue;
320 		}
321 		/*
322 		 * merge overlapping changes in first file
323 		 * this happens after extension see below
324 		 */
325 		if (d1+1 < d13+m1 && d1->new.to >= d1[1].new.from) {
326 			d1[1].old.from = d1->old.from;
327 			d1[1].new.from = d1->new.from;
328 			d1++;
329 			continue;
330 		}
331 
332 		/* merge overlapping changes in second */
333 		if (d2+1 < d23+m2 && d2->new.to >= d2[1].new.from) {
334 			d2[1].old.from = d2->old.from;
335 			d2[1].new.from = d2->new.from;
336 			d2++;
337 			continue;
338 		}
339 
340 		/* stuff peculiar to third file or different in all */
341 		if (d1->new.from == d2->new.from && d1->new.to == d2->new.to) {
342 			dup = duplicate(&d1->old, &d2->old);
343 			/*
344 			 * dup = 0 means all files differ
345 			 * dup = 1 meands files 1&2 identical
346 			 */
347 			if (eflag == 0) {
348 				separate(dup?"3":"");
349 				change(1, &d1->old, dup);
350 				change(2, &d2->old, 0);
351 				d3 = d1->old.to > d1->old.from ? d1 : d2;
352 				change(3, &d3->new, 0);
353 			} else
354 				j = edit(d1, dup, j);
355 			d1++;
356 			d2++;
357 			continue;
358 		}
359 		/*
360 		 * overlapping changes from file1 & 2
361 		 * extend changes appropriately to
362 		 * make them coincide
363 		 */
364 		if (d1->new.from < d2->new.from) {
365 			d2->old.from -= d2->new.from-d1->new.from;
366 			d2->new.from = d1->new.from;
367 		} else if (d2->new.from < d1->new.from) {
368 			d1->old.from -= d1->new.from-d2->new.from;
369 			d1->new.from = d2->new.from;
370 		}
371 
372 		if (d1->new.to > d2->new.to) {
373 			d2->old.to += d1->new.to - d2->new.to;
374 			d2->new.to = d1->new.to;
375 		} else if (d2->new.to > d1->new.to) {
376 			d1->old.to += d2->new.to - d1->new.to;
377 			d1->new.to = d2->new.to;
378 		}
379 	}
380 	if (eflag) {
381 		edscript(j);
382 		if (j)
383 			(void) printf("w\nq\n");
384 	}
385 }
386 
387 static void
388 separate(char *s)
389 {
390 	(void) printf("====%s\n", s);
391 }
392 
393 /*
394  * the range of ines rold.from thru rold.to in file i
395  * is to be changed. it is to be printed only if
396  * it does not duplicate something to be printed later
397  */
398 static void
399 change(int i, struct range *rold, int dup)
400 {
401 	(void) printf("%d:", i);
402 	last[i] = rold->to;
403 	prange(rold);
404 	if (dup)
405 		return;
406 	if (debug)
407 		return;
408 	i--;
409 	(void) skip(i, rold->from, (char *)0);
410 	(void) skip(i, rold->to, "  ");
411 }
412 
413 /*
414  * print the range of line numbers, rold.from  thru rold.to
415  * as n1, n2 or n1
416  */
417 static void
418 prange(struct range *rold)
419 {
420 	if (rold->to <= rold->from)
421 		(void) printf("%da\n", rold->from-1);
422 	else {
423 		(void) printf("%d", rold->from);
424 		if (rold->to > rold->from+1)
425 			(void) printf(",%d", rold->to-1);
426 		(void) printf("c\n");
427 	}
428 }
429 
430 /*
431  * no difference was reported by diff between file 1(or 2)
432  * and file 3, and an artificial dummy difference (trange)
433  * must be ginned up to correspond to the change reported
434  * in the other file
435  */
436 static void
437 keep(int i, struct range *rnew)
438 {
439 	int delta;
440 	struct range trange;
441 	delta = last[3] - last[i];
442 	trange.from = rnew->from - delta;
443 	trange.to = rnew->to - delta;
444 	change(i, &trange, 1);
445 }
446 
447 /*
448  * skip to just befor line number from in file i
449  * if "pr" is nonzero, print all skipped stuff
450  * with string pr as a prefix
451  */
452 static int
453 skip(int i, int from, char *pr)
454 {
455 	int j, n;
456 	for (n = 0; cline[i] < from-1; n += j) {
457 		if ((j = getline(fp[i])) == 0)
458 			trouble();
459 		if (pr)
460 			(void) printf("%s%s", pr, line);
461 		cline[i]++;
462 	}
463 	return (n);
464 }
465 
466 /*
467  * return 1 or 0 according as the old range
468  * (in file 1) contains exactly the same data
469  * as the new range (in file 2)
470  */
471 static int
472 duplicate(struct range *r1, struct range *r2)
473 {
474 	int c, d;
475 	int nchar;
476 	int nline;
477 	if (r1->to-r1->from != r2->to-r2->from)
478 		return (0);
479 	(void) skip(0, r1->from, (char *)0);
480 	(void) skip(1, r2->from, (char *)0);
481 	nchar = 0;
482 	for (nline = 0; nline < r1->to-r1->from; nline++) {
483 		do {
484 			c = getc(fp[0]);
485 			d = getc(fp[1]);
486 			if (c == -1 || d == -1)
487 				trouble();
488 			nchar++;
489 			if (c != d) {
490 				repos(nchar);
491 				return (0);
492 			}
493 		} while (c != '\n');
494 	}
495 	repos(nchar);
496 	return (1);
497 }
498 
499 static void
500 repos(int nchar)
501 {
502 	int i;
503 	for (i = 0; i < 2; i++)
504 		(void) fseek(fp[i], (long)-nchar, 1);
505 }
506 
507 static void
508 trouble()
509 {
510 	(void) fprintf(stderr, "diff3: logic error\n");
511 	abort();
512 }
513 
514 /*
515  * collect an editing script for later regurgitation
516  */
517 static int
518 edit(struct diff *diff, int dup, int j)
519 {
520 	if (((dup+1)&eflag) == 0)
521 		return (j);
522 	j++;
523 	overlap[j] = !dup;
524 	if (!dup) overlapcnt++;
525 	de[j].old.from = diff->old.from;
526 	de[j].old.to = diff->old.to;
527 	de[j].new.from = de[j-1].new.to + skip(2, diff->new.from, (char *)0);
528 	de[j].new.to = de[j].new.from + skip(2, diff->new.to, (char *)0);
529 	return (j);
530 }
531 
532 /*		regurgitate */
533 static void
534 edscript(int n)
535 {
536 	int j, k;
537 	char	 block[BUFSIZ];
538 
539 	for (n = n; n > 0; n--) {
540 		if (!oflag || !overlap[n])
541 			prange(&de[n].old);
542 		else
543 			(void) printf("%da\n=======\n", de[n].old.to -1);
544 		(void) fseek(fp[2], (long)de[n].new.from, 0);
545 		for (k = de[n].new.to-de[n].new.from; k > 0; k -= j) {
546 			j = k > BUFSIZ?BUFSIZ:k;
547 			if (fread(block, 1, j, fp[2]) != j)
548 				trouble();
549 			(void) fwrite(block, 1, j, stdout);
550 		}
551 		if (!oflag || !overlap[n])
552 			(void) printf(".\n");
553 		else {
554 			(void) printf("%s\n.\n", f3mark);
555 			(void) printf("%da\n%s\n.\n", de[n].old.from-1, f1mark);
556 		}
557 	}
558 }
559 
560 static void
561 usage()
562 {
563 	(void) fprintf(stderr,
564 	    "\tusage: diff3prog [-ex3EX] d13 d23 f1 f2 f3 [m1 m2]\n");
565 	exit(1);
566 }
567