xref: /openbsd-src/usr.bin/diff/diff.c (revision a28daedfc357b214be5c701aa8ba8adb29a7f1c2)
1 /*	$OpenBSD: diff.c,v 1.50 2007/05/29 18:24:56 ray Exp $	*/
2 
3 /*
4  * Copyright (c) 2003 Todd C. Miller <Todd.Miller@courtesan.com>
5  *
6  * Permission to use, copy, modify, and distribute this software for any
7  * purpose with or without fee is hereby granted, provided that the above
8  * copyright notice and this permission notice appear in all copies.
9  *
10  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17  *
18  * Sponsored in part by the Defense Advanced Research Projects
19  * Agency (DARPA) and Air Force Research Laboratory, Air Force
20  * Materiel Command, USAF, under agreement number F39502-99-1-0512.
21  */
22 
23 #ifndef lint
24 static const char rcsid[] = "$OpenBSD: diff.c,v 1.50 2007/05/29 18:24:56 ray Exp $";
25 #endif /* not lint */
26 
27 #include <sys/param.h>
28 #include <sys/stat.h>
29 
30 #include <ctype.h>
31 #include <err.h>
32 #include <errno.h>
33 #include <getopt.h>
34 #include <signal.h>
35 #include <stdlib.h>
36 #include <stdio.h>
37 #include <stdarg.h>
38 #include <string.h>
39 #include <unistd.h>
40 
41 #include "diff.h"
42 #include "xmalloc.h"
43 
44 int	 aflag, bflag, dflag, iflag, lflag, Nflag, Pflag, pflag, rflag;
45 int	 sflag, tflag, Tflag, wflag;
46 int	 format, context, status;
47 char	*start, *ifdefname, *diffargs, *label[2], *ignore_pats;
48 struct stat stb1, stb2;
49 struct excludes *excludes_list;
50 regex_t	 ignore_re;
51 
52 #define	OPTIONS	"0123456789abC:cdD:efhI:iL:lnNPpqrS:sTtU:uwX:x:"
53 static struct option longopts[] = {
54 	{ "text",			no_argument,		0,	'a' },
55 	{ "ignore-space-change",	no_argument,		0,	'b' },
56 	{ "context",			optional_argument,	0,	'C' },
57 	{ "ifdef",			required_argument,	0,	'D' },
58 	{ "minimal",			no_argument,		0,	'd' },
59 	{ "ed",				no_argument,		0,	'e' },
60 	{ "forward-ed",			no_argument,		0,	'f' },
61 	{ "ignore-matching-lines",	required_argument,	0,	'I' },
62 	{ "ignore-case",		no_argument,		0,	'i' },
63 	{ "paginate",			no_argument,		0,	'l' },
64 	{ "label",			required_argument,	0,	'L' },
65 	{ "new-file",			no_argument,		0,	'N' },
66 	{ "rcs",			no_argument,		0,	'n' },
67 	{ "unidirectional-new-file",	no_argument,		0,	'P' },
68 	{ "show-c-function",		no_argument,		0,	'p' },
69 	{ "brief",			no_argument,		0,	'q' },
70 	{ "recursive",			no_argument,		0,	'r' },
71 	{ "report-identical-files",	no_argument,		0,	's' },
72 	{ "starting-file",		required_argument,	0,	'S' },
73 	{ "expand-tabs",		no_argument,		0,	't' },
74 	{ "initial-tab",		no_argument,		0,	'T' },
75 	{ "unified",			optional_argument,	0,	'U' },
76 	{ "ignore-all-space",		no_argument,		0,	'w' },
77 	{ "exclude",			required_argument,	0,	'x' },
78 	{ "exclude-from",		required_argument,	0,	'X' },
79 	{ NULL,				0,			0,	'\0'}
80 };
81 
82 __dead void usage(void);
83 void push_excludes(char *);
84 void push_ignore_pats(char *);
85 void read_excludes_file(char *file);
86 void set_argstr(char **, char **);
87 
88 int
89 main(int argc, char **argv)
90 {
91 	char *ep, **oargv;
92 	long  l;
93 	int   ch, lastch, gotstdin, prevoptind, newarg;
94 
95 	oargv = argv;
96 	gotstdin = 0;
97 
98 	lastch = '\0';
99 	prevoptind = 1;
100 	newarg = 1;
101 	while ((ch = getopt_long(argc, argv, OPTIONS, longopts, NULL)) != -1) {
102 		switch (ch) {
103 		case '0': case '1': case '2': case '3': case '4':
104 		case '5': case '6': case '7': case '8': case '9':
105 			if (newarg)
106 				usage();	/* disallow -[0-9]+ */
107 			else if (lastch == 'c' || lastch == 'u')
108 				context = 0;
109 			else if (!isdigit(lastch) || context > INT_MAX / 10)
110 				usage();
111 			context = (context * 10) + (ch - '0');
112 			break;
113 		case 'a':
114 			aflag = 1;
115 			break;
116 		case 'b':
117 			bflag = 1;
118 			break;
119 		case 'C':
120 		case 'c':
121 			format = D_CONTEXT;
122 			if (optarg != NULL) {
123 				l = strtol(optarg, &ep, 10);
124 				if (*ep != '\0' || l < 0 || l >= INT_MAX)
125 					usage();
126 				context = (int)l;
127 			} else
128 				context = 3;
129 			break;
130 		case 'd':
131 			dflag = 1;
132 			break;
133 		case 'D':
134 			format = D_IFDEF;
135 			ifdefname = optarg;
136 			break;
137 		case 'e':
138 			format = D_EDIT;
139 			break;
140 		case 'f':
141 			format = D_REVERSE;
142 			break;
143 		case 'h':
144 			/* silently ignore for backwards compatibility */
145 			break;
146 		case 'I':
147 			push_ignore_pats(optarg);
148 			break;
149 		case 'i':
150 			iflag = 1;
151 			break;
152 		case 'L':
153 			if (label[0] == NULL)
154 				label[0] = optarg;
155 			else if (label[1] == NULL)
156 				label[1] = optarg;
157 			else
158 				usage();
159 			break;
160 		case 'l':
161 			lflag = 1;
162 			signal(SIGPIPE, SIG_IGN);
163 			break;
164 		case 'N':
165 			Nflag = 1;
166 			break;
167 		case 'n':
168 			format = D_NREVERSE;
169 			break;
170 		case 'p':
171 			pflag = 1;
172 			break;
173 		case 'P':
174 			Pflag = 1;
175 			break;
176 		case 'r':
177 			rflag = 1;
178 			break;
179 		case 'q':
180 			format = D_BRIEF;
181 			break;
182 		case 'S':
183 			start = optarg;
184 			break;
185 		case 's':
186 			sflag = 1;
187 			break;
188 		case 'T':
189 			Tflag = 1;
190 			break;
191 		case 't':
192 			tflag = 1;
193 			break;
194 		case 'U':
195 		case 'u':
196 			format = D_UNIFIED;
197 			if (optarg != NULL) {
198 				l = strtol(optarg, &ep, 10);
199 				if (*ep != '\0' || l < 0 || l >= INT_MAX)
200 					usage();
201 				context = (int)l;
202 			} else
203 				context = 3;
204 			break;
205 		case 'w':
206 			wflag = 1;
207 			break;
208 		case 'X':
209 			read_excludes_file(optarg);
210 			break;
211 		case 'x':
212 			push_excludes(optarg);
213 			break;
214 		default:
215 			usage();
216 			break;
217 		}
218 		lastch = ch;
219 		newarg = optind != prevoptind;
220 		prevoptind = optind;
221 	}
222 	argc -= optind;
223 	argv += optind;
224 
225 	/*
226 	 * Do sanity checks, fill in stb1 and stb2 and call the appropriate
227 	 * driver routine.  Both drivers use the contents of stb1 and stb2.
228 	 */
229 	if (argc != 2)
230 		usage();
231 	if (ignore_pats != NULL) {
232 		char buf[BUFSIZ];
233 		int error;
234 
235 		if ((error = regcomp(&ignore_re, ignore_pats,
236 				     REG_NEWLINE | REG_EXTENDED)) != 0) {
237 			regerror(error, &ignore_re, buf, sizeof(buf));
238 			if (*ignore_pats != '\0')
239 				errx(2, "%s: %s", ignore_pats, buf);
240 			else
241 				errx(2, "%s", buf);
242 		}
243 	}
244 	if (strcmp(argv[0], "-") == 0) {
245 		fstat(STDIN_FILENO, &stb1);
246 		gotstdin = 1;
247 	} else if (stat(argv[0], &stb1) != 0)
248 		err(2, "%s", argv[0]);
249 	if (strcmp(argv[1], "-") == 0) {
250 		fstat(STDIN_FILENO, &stb2);
251 		gotstdin = 1;
252 	} else if (stat(argv[1], &stb2) != 0)
253 		err(2, "%s", argv[1]);
254 	if (gotstdin && (S_ISDIR(stb1.st_mode) || S_ISDIR(stb2.st_mode)))
255 		errx(2, "can't compare - to a directory");
256 	set_argstr(oargv, argv);
257 	if (S_ISDIR(stb1.st_mode) && S_ISDIR(stb2.st_mode)) {
258 		if (format == D_IFDEF)
259 			errx(2, "-D option not supported with directories");
260 		diffdir(argv[0], argv[1]);
261 	} else {
262 		if (S_ISDIR(stb1.st_mode)) {
263 			argv[0] = splice(argv[0], argv[1]);
264 			if (stat(argv[0], &stb1) < 0)
265 				err(2, "%s", argv[0]);
266 		}
267 		if (S_ISDIR(stb2.st_mode)) {
268 			argv[1] = splice(argv[1], argv[0]);
269 			if (stat(argv[1], &stb2) < 0)
270 				err(2, "%s", argv[1]);
271 		}
272 		print_status(diffreg(argv[0], argv[1], 0), argv[0], argv[1],
273 		    NULL);
274 	}
275 	exit(status);
276 }
277 
278 void
279 set_argstr(char **av, char **ave)
280 {
281 	size_t argsize;
282 	char **ap;
283 
284 	argsize = 4 + *ave - *av + 1;
285 	diffargs = xmalloc(argsize);
286 	strlcpy(diffargs, "diff", argsize);
287 	for (ap = av + 1; ap < ave; ap++) {
288 		if (strcmp(*ap, "--") != 0) {
289 			strlcat(diffargs, " ", argsize);
290 			strlcat(diffargs, *ap, argsize);
291 		}
292 	}
293 }
294 
295 /*
296  * Read in an excludes file and push each line.
297  */
298 void
299 read_excludes_file(char *file)
300 {
301 	FILE *fp;
302 	char *buf, *pattern;
303 	size_t len;
304 
305 	if (strcmp(file, "-") == 0)
306 		fp = stdin;
307 	else if ((fp = fopen(file, "r")) == NULL)
308 		err(2, "%s", file);
309 	while ((buf = fgetln(fp, &len)) != NULL) {
310 		if (buf[len - 1] == '\n')
311 			len--;
312 		pattern = xmalloc(len + 1);
313 		memcpy(pattern, buf, len);
314 		pattern[len] = '\0';
315 		push_excludes(pattern);
316 	}
317 	if (strcmp(file, "-") != 0)
318 		fclose(fp);
319 }
320 
321 /*
322  * Push a pattern onto the excludes list.
323  */
324 void
325 push_excludes(char *pattern)
326 {
327 	struct excludes *entry;
328 
329 	entry = xmalloc(sizeof(*entry));
330 	entry->pattern = pattern;
331 	entry->next = excludes_list;
332 	excludes_list = entry;
333 }
334 
335 void
336 push_ignore_pats(char *pattern)
337 {
338 	size_t len;
339 
340 	if (ignore_pats == NULL)
341 		ignore_pats = xstrdup(pattern);
342 	else {
343 		/* old + "|" + new + NUL */
344 		len = strlen(ignore_pats) + strlen(pattern) + 2;
345 		ignore_pats = xrealloc(ignore_pats, 1, len);
346 		strlcat(ignore_pats, "|", len);
347 		strlcat(ignore_pats, pattern, len);
348 	}
349 }
350 
351 void
352 print_only(const char *path, size_t dirlen, const char *entry)
353 {
354 	if (dirlen > 1)
355 		dirlen--;
356 	printf("Only in %.*s: %s\n", (int)dirlen, path, entry);
357 }
358 
359 void
360 print_status(int val, char *path1, char *path2, char *entry)
361 {
362 	switch (val) {
363 	case D_ONLY:
364 		print_only(path1, strlen(path1), entry);
365 		break;
366 	case D_COMMON:
367 		printf("Common subdirectories: %s%s and %s%s\n",
368 		    path1, entry ? entry : "", path2, entry ? entry : "");
369 		break;
370 	case D_BINARY:
371 		printf("Binary files %s%s and %s%s differ\n",
372 		    path1, entry ? entry : "", path2, entry ? entry : "");
373 		break;
374 	case D_DIFFER:
375 		if (format == D_BRIEF)
376 			printf("Files %s%s and %s%s differ\n",
377 			    path1, entry ? entry : "",
378 			    path2, entry ? entry : "");
379 		break;
380 	case D_SAME:
381 		if (sflag)
382 			printf("Files %s%s and %s%s are identical\n",
383 			    path1, entry ? entry : "",
384 			    path2, entry ? entry : "");
385 		break;
386 	case D_MISMATCH1:
387 		printf("File %s%s is a directory while file %s%s is a regular file\n",
388 		    path1, entry ? entry : "", path2, entry ? entry : "");
389 		break;
390 	case D_MISMATCH2:
391 		printf("File %s%s is a regular file while file %s%s is a directory\n",
392 		    path1, entry ? entry : "", path2, entry ? entry : "");
393 		break;
394 	case D_SKIPPED1:
395 		printf("File %s%s is not a regular file or directory and was skipped\n",
396 		    path1, entry ? entry : "");
397 		break;
398 	case D_SKIPPED2:
399 		printf("File %s%s is not a regular file or directory and was skipped\n",
400 		    path2, entry ? entry : "");
401 		break;
402 	}
403 }
404 
405 __dead void
406 usage(void)
407 {
408 	(void)fprintf(stderr,
409 	    "usage: diff [-abdilpqTtw] [-I pattern] [-c | -e | -f | -n | -u]\n"
410 	    "            [-L label] file1 file2\n"
411 	    "       diff [-abdilpqTtw] [-I pattern] [-L label] -C number file1 file2\n"
412 	    "       diff [-abdilqtw] [-I pattern] -D string file1 file2\n"
413 	    "       diff [-abdilpqTtw] [-I pattern] [-L label] -U number file1 file2\n"
414 	    "       diff [-abdilNPpqrsTtw] [-I pattern] [-c | -e | -f | -n | -u]\n"
415 	    "            [-L label] [-S name] [-X file] [-x pattern] dir1 dir2\n");
416 
417 	exit(2);
418 }
419