xref: /openbsd-src/usr.bin/diff/diff.c (revision dba1d6ea0e7adc7407ba8b634ca533c869704ccd)
1 /*	$OpenBSD: diff.c,v 1.51 2009/06/06 15:00:27 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 1.51 2009/06/06 15:00:27 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	 lflag, Nflag, Pflag, rflag, sflag, Tflag;
45 int	 format, context, status;
46 char	*start, *ifdefname, *diffargs, *label[2], *ignore_pats;
47 struct stat stb1, stb2;
48 struct excludes *excludes_list;
49 regex_t	 ignore_re;
50 
51 #define	OPTIONS	"0123456789abC:cdD:efhI:iL:lnNPpqrS:sTtU:uwX:x:"
52 static struct option longopts[] = {
53 	{ "text",			no_argument,		0,	'a' },
54 	{ "ignore-space-change",	no_argument,		0,	'b' },
55 	{ "context",			optional_argument,	0,	'C' },
56 	{ "ifdef",			required_argument,	0,	'D' },
57 	{ "minimal",			no_argument,		0,	'd' },
58 	{ "ed",				no_argument,		0,	'e' },
59 	{ "forward-ed",			no_argument,		0,	'f' },
60 	{ "ignore-matching-lines",	required_argument,	0,	'I' },
61 	{ "ignore-case",		no_argument,		0,	'i' },
62 	{ "paginate",			no_argument,		0,	'l' },
63 	{ "label",			required_argument,	0,	'L' },
64 	{ "new-file",			no_argument,		0,	'N' },
65 	{ "rcs",			no_argument,		0,	'n' },
66 	{ "unidirectional-new-file",	no_argument,		0,	'P' },
67 	{ "show-c-function",		no_argument,		0,	'p' },
68 	{ "brief",			no_argument,		0,	'q' },
69 	{ "recursive",			no_argument,		0,	'r' },
70 	{ "report-identical-files",	no_argument,		0,	's' },
71 	{ "starting-file",		required_argument,	0,	'S' },
72 	{ "expand-tabs",		no_argument,		0,	't' },
73 	{ "initial-tab",		no_argument,		0,	'T' },
74 	{ "unified",			optional_argument,	0,	'U' },
75 	{ "ignore-all-space",		no_argument,		0,	'w' },
76 	{ "exclude",			required_argument,	0,	'x' },
77 	{ "exclude-from",		required_argument,	0,	'X' },
78 	{ NULL,				0,			0,	'\0'}
79 };
80 
81 __dead void usage(void);
82 void push_excludes(char *);
83 void push_ignore_pats(char *);
84 void read_excludes_file(char *file);
85 void set_argstr(char **, char **);
86 
87 int
88 main(int argc, char **argv)
89 {
90 	char *ep, **oargv;
91 	long  l;
92 	int   ch, dflags, lastch, gotstdin, prevoptind, newarg;
93 
94 	oargv = argv;
95 	gotstdin = 0;
96 	dflags = 0;
97 	lastch = '\0';
98 	prevoptind = 1;
99 	newarg = 1;
100 	while ((ch = getopt_long(argc, argv, OPTIONS, longopts, NULL)) != -1) {
101 		switch (ch) {
102 		case '0': case '1': case '2': case '3': case '4':
103 		case '5': case '6': case '7': case '8': case '9':
104 			if (newarg)
105 				usage();	/* disallow -[0-9]+ */
106 			else if (lastch == 'c' || lastch == 'u')
107 				context = 0;
108 			else if (!isdigit(lastch) || context > INT_MAX / 10)
109 				usage();
110 			context = (context * 10) + (ch - '0');
111 			break;
112 		case 'a':
113 			dflags |= D_FORCEASCII;
114 			break;
115 		case 'b':
116 			dflags |= D_FOLDBLANKS;
117 			break;
118 		case 'C':
119 		case 'c':
120 			format = D_CONTEXT;
121 			if (optarg != NULL) {
122 				l = strtol(optarg, &ep, 10);
123 				if (*ep != '\0' || l < 0 || l >= INT_MAX)
124 					usage();
125 				context = (int)l;
126 			} else
127 				context = 3;
128 			break;
129 		case 'd':
130 			dflags |= D_MINIMAL;
131 			break;
132 		case 'D':
133 			format = D_IFDEF;
134 			ifdefname = optarg;
135 			break;
136 		case 'e':
137 			format = D_EDIT;
138 			break;
139 		case 'f':
140 			format = D_REVERSE;
141 			break;
142 		case 'h':
143 			/* silently ignore for backwards compatibility */
144 			break;
145 		case 'I':
146 			push_ignore_pats(optarg);
147 			break;
148 		case 'i':
149 			dflags |= D_IGNORECASE;
150 			break;
151 		case 'L':
152 			if (label[0] == NULL)
153 				label[0] = optarg;
154 			else if (label[1] == NULL)
155 				label[1] = optarg;
156 			else
157 				usage();
158 			break;
159 		case 'l':
160 			lflag = 1;
161 			signal(SIGPIPE, SIG_IGN);
162 			break;
163 		case 'N':
164 			Nflag = 1;
165 			break;
166 		case 'n':
167 			format = D_NREVERSE;
168 			break;
169 		case 'p':
170 			dflags |= D_PROTOTYPE;
171 			break;
172 		case 'P':
173 			Pflag = 1;
174 			break;
175 		case 'r':
176 			rflag = 1;
177 			break;
178 		case 'q':
179 			format = D_BRIEF;
180 			break;
181 		case 'S':
182 			start = optarg;
183 			break;
184 		case 's':
185 			sflag = 1;
186 			break;
187 		case 'T':
188 			Tflag = 1;
189 			break;
190 		case 't':
191 			dflags |= D_EXPANDTABS;
192 			break;
193 		case 'U':
194 		case 'u':
195 			format = D_UNIFIED;
196 			if (optarg != NULL) {
197 				l = strtol(optarg, &ep, 10);
198 				if (*ep != '\0' || l < 0 || l >= INT_MAX)
199 					usage();
200 				context = (int)l;
201 			} else
202 				context = 3;
203 			break;
204 		case 'w':
205 			dflags |= D_IGNOREBLANKS;
206 			break;
207 		case 'X':
208 			read_excludes_file(optarg);
209 			break;
210 		case 'x':
211 			push_excludes(optarg);
212 			break;
213 		default:
214 			usage();
215 			break;
216 		}
217 		lastch = ch;
218 		newarg = optind != prevoptind;
219 		prevoptind = optind;
220 	}
221 	argc -= optind;
222 	argv += optind;
223 
224 	/*
225 	 * Do sanity checks, fill in stb1 and stb2 and call the appropriate
226 	 * driver routine.  Both drivers use the contents of stb1 and stb2.
227 	 */
228 	if (argc != 2)
229 		usage();
230 	if (ignore_pats != NULL) {
231 		char buf[BUFSIZ];
232 		int error;
233 
234 		if ((error = regcomp(&ignore_re, ignore_pats,
235 				     REG_NEWLINE | REG_EXTENDED)) != 0) {
236 			regerror(error, &ignore_re, buf, sizeof(buf));
237 			if (*ignore_pats != '\0')
238 				errx(2, "%s: %s", ignore_pats, buf);
239 			else
240 				errx(2, "%s", buf);
241 		}
242 	}
243 	if (strcmp(argv[0], "-") == 0) {
244 		fstat(STDIN_FILENO, &stb1);
245 		gotstdin = 1;
246 	} else if (stat(argv[0], &stb1) != 0)
247 		err(2, "%s", argv[0]);
248 	if (strcmp(argv[1], "-") == 0) {
249 		fstat(STDIN_FILENO, &stb2);
250 		gotstdin = 1;
251 	} else if (stat(argv[1], &stb2) != 0)
252 		err(2, "%s", argv[1]);
253 	if (gotstdin && (S_ISDIR(stb1.st_mode) || S_ISDIR(stb2.st_mode)))
254 		errx(2, "can't compare - to a directory");
255 	set_argstr(oargv, argv);
256 	if (S_ISDIR(stb1.st_mode) && S_ISDIR(stb2.st_mode)) {
257 		if (format == D_IFDEF)
258 			errx(2, "-D option not supported with directories");
259 		diffdir(argv[0], argv[1]);
260 	} else {
261 		if (S_ISDIR(stb1.st_mode)) {
262 			argv[0] = splice(argv[0], argv[1]);
263 			if (stat(argv[0], &stb1) < 0)
264 				err(2, "%s", argv[0]);
265 		}
266 		if (S_ISDIR(stb2.st_mode)) {
267 			argv[1] = splice(argv[1], argv[0]);
268 			if (stat(argv[1], &stb2) < 0)
269 				err(2, "%s", argv[1]);
270 		}
271 		print_status(diffreg(argv[0], argv[1], 0), argv[0], argv[1],
272 		    NULL);
273 	}
274 	exit(status);
275 }
276 
277 void
278 set_argstr(char **av, char **ave)
279 {
280 	size_t argsize;
281 	char **ap;
282 
283 	argsize = 4 + *ave - *av + 1;
284 	diffargs = xmalloc(argsize);
285 	strlcpy(diffargs, "diff", argsize);
286 	for (ap = av + 1; ap < ave; ap++) {
287 		if (strcmp(*ap, "--") != 0) {
288 			strlcat(diffargs, " ", argsize);
289 			strlcat(diffargs, *ap, argsize);
290 		}
291 	}
292 }
293 
294 /*
295  * Read in an excludes file and push each line.
296  */
297 void
298 read_excludes_file(char *file)
299 {
300 	FILE *fp;
301 	char *buf, *pattern;
302 	size_t len;
303 
304 	if (strcmp(file, "-") == 0)
305 		fp = stdin;
306 	else if ((fp = fopen(file, "r")) == NULL)
307 		err(2, "%s", file);
308 	while ((buf = fgetln(fp, &len)) != NULL) {
309 		if (buf[len - 1] == '\n')
310 			len--;
311 		pattern = xmalloc(len + 1);
312 		memcpy(pattern, buf, len);
313 		pattern[len] = '\0';
314 		push_excludes(pattern);
315 	}
316 	if (strcmp(file, "-") != 0)
317 		fclose(fp);
318 }
319 
320 /*
321  * Push a pattern onto the excludes list.
322  */
323 void
324 push_excludes(char *pattern)
325 {
326 	struct excludes *entry;
327 
328 	entry = xmalloc(sizeof(*entry));
329 	entry->pattern = pattern;
330 	entry->next = excludes_list;
331 	excludes_list = entry;
332 }
333 
334 void
335 push_ignore_pats(char *pattern)
336 {
337 	size_t len;
338 
339 	if (ignore_pats == NULL)
340 		ignore_pats = xstrdup(pattern);
341 	else {
342 		/* old + "|" + new + NUL */
343 		len = strlen(ignore_pats) + strlen(pattern) + 2;
344 		ignore_pats = xrealloc(ignore_pats, 1, len);
345 		strlcat(ignore_pats, "|", len);
346 		strlcat(ignore_pats, pattern, len);
347 	}
348 }
349 
350 void
351 print_only(const char *path, size_t dirlen, const char *entry)
352 {
353 	if (dirlen > 1)
354 		dirlen--;
355 	printf("Only in %.*s: %s\n", (int)dirlen, path, entry);
356 }
357 
358 void
359 print_status(int val, char *path1, char *path2, char *entry)
360 {
361 	switch (val) {
362 	case D_ONLY:
363 		print_only(path1, strlen(path1), entry);
364 		break;
365 	case D_COMMON:
366 		printf("Common subdirectories: %s%s and %s%s\n",
367 		    path1, entry ? entry : "", path2, entry ? entry : "");
368 		break;
369 	case D_BINARY:
370 		printf("Binary files %s%s and %s%s differ\n",
371 		    path1, entry ? entry : "", path2, entry ? entry : "");
372 		break;
373 	case D_DIFFER:
374 		if (format == D_BRIEF)
375 			printf("Files %s%s and %s%s differ\n",
376 			    path1, entry ? entry : "",
377 			    path2, entry ? entry : "");
378 		break;
379 	case D_SAME:
380 		if (sflag)
381 			printf("Files %s%s and %s%s are identical\n",
382 			    path1, entry ? entry : "",
383 			    path2, entry ? entry : "");
384 		break;
385 	case D_MISMATCH1:
386 		printf("File %s%s is a directory while file %s%s is a regular file\n",
387 		    path1, entry ? entry : "", path2, entry ? entry : "");
388 		break;
389 	case D_MISMATCH2:
390 		printf("File %s%s is a regular file while file %s%s is a directory\n",
391 		    path1, entry ? entry : "", path2, entry ? entry : "");
392 		break;
393 	case D_SKIPPED1:
394 		printf("File %s%s is not a regular file or directory and was skipped\n",
395 		    path1, entry ? entry : "");
396 		break;
397 	case D_SKIPPED2:
398 		printf("File %s%s is not a regular file or directory and was skipped\n",
399 		    path2, entry ? entry : "");
400 		break;
401 	}
402 }
403 
404 __dead void
405 usage(void)
406 {
407 	(void)fprintf(stderr,
408 	    "usage: diff [-abdilpqTtw] [-I pattern] [-c | -e | -f | -n | -u]\n"
409 	    "            [-L label] file1 file2\n"
410 	    "       diff [-abdilpqTtw] [-I pattern] [-L label] -C number file1 file2\n"
411 	    "       diff [-abdilqtw] [-I pattern] -D string file1 file2\n"
412 	    "       diff [-abdilpqTtw] [-I pattern] [-L label] -U number file1 file2\n"
413 	    "       diff [-abdilNPpqrsTtw] [-I pattern] [-c | -e | -f | -n | -u]\n"
414 	    "            [-L label] [-S name] [-X file] [-x pattern] dir1 dir2\n");
415 
416 	exit(2);
417 }
418