xref: /openbsd/usr.bin/diff/diff.c (revision 771fbea0)
1 /*	$OpenBSD: diff.c,v 1.67 2019/06/28 13:35:00 deraadt Exp $	*/
2 
3 /*
4  * Copyright (c) 2003 Todd C. Miller <millert@openbsd.org>
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 #include <sys/stat.h>
24 
25 #include <ctype.h>
26 #include <err.h>
27 #include <errno.h>
28 #include <getopt.h>
29 #include <stdlib.h>
30 #include <stdio.h>
31 #include <stdarg.h>
32 #include <string.h>
33 #include <unistd.h>
34 #include <limits.h>
35 
36 #include "diff.h"
37 #include "xmalloc.h"
38 
39 int	 Nflag, Pflag, rflag, sflag, Tflag;
40 int	 diff_format, diff_context, status;
41 char	*start, *ifdefname, *diffargs, *label[2], *ignore_pats;
42 struct stat stb1, stb2;
43 struct excludes *excludes_list;
44 regex_t	 ignore_re;
45 
46 #define	OPTIONS	"0123456789abC:cdD:efhI:iL:lnNPpqrS:sTtU:uwX:x:"
47 static struct option longopts[] = {
48 	{ "text",			no_argument,		0,	'a' },
49 	{ "ignore-space-change",	no_argument,		0,	'b' },
50 	{ "context",			optional_argument,	0,	'C' },
51 	{ "ifdef",			required_argument,	0,	'D' },
52 	{ "minimal",			no_argument,		0,	'd' },
53 	{ "ed",				no_argument,		0,	'e' },
54 	{ "forward-ed",			no_argument,		0,	'f' },
55 	{ "ignore-matching-lines",	required_argument,	0,	'I' },
56 	{ "ignore-case",		no_argument,		0,	'i' },
57 	{ "label",			required_argument,	0,	'L' },
58 	{ "new-file",			no_argument,		0,	'N' },
59 	{ "rcs",			no_argument,		0,	'n' },
60 	{ "unidirectional-new-file",	no_argument,		0,	'P' },
61 	{ "show-c-function",		no_argument,		0,	'p' },
62 	{ "brief",			no_argument,		0,	'q' },
63 	{ "recursive",			no_argument,		0,	'r' },
64 	{ "report-identical-files",	no_argument,		0,	's' },
65 	{ "starting-file",		required_argument,	0,	'S' },
66 	{ "expand-tabs",		no_argument,		0,	't' },
67 	{ "initial-tab",		no_argument,		0,	'T' },
68 	{ "unified",			optional_argument,	0,	'U' },
69 	{ "ignore-all-space",		no_argument,		0,	'w' },
70 	{ "exclude",			required_argument,	0,	'x' },
71 	{ "exclude-from",		required_argument,	0,	'X' },
72 	{ NULL,				0,			0,	'\0'}
73 };
74 
75 __dead void usage(void);
76 void push_excludes(char *);
77 void push_ignore_pats(char *);
78 void read_excludes_file(char *file);
79 void set_argstr(char **, char **);
80 
81 int
82 main(int argc, char **argv)
83 {
84 	char *ep, **oargv;
85 	long  l;
86 	int   ch, dflags, lastch, gotstdin, prevoptind, newarg;
87 
88 	oargv = argv;
89 	gotstdin = 0;
90 	dflags = 0;
91 	lastch = '\0';
92 	prevoptind = 1;
93 	newarg = 1;
94 	while ((ch = getopt_long(argc, argv, OPTIONS, longopts, NULL)) != -1) {
95 		switch (ch) {
96 		case '0': case '1': case '2': case '3': case '4':
97 		case '5': case '6': case '7': case '8': case '9':
98 			if (newarg)
99 				usage();	/* disallow -[0-9]+ */
100 			else if (lastch == 'c' || lastch == 'u')
101 				diff_context = 0;
102 			else if (!isdigit(lastch) || diff_context > INT_MAX / 10)
103 				usage();
104 			diff_context = (diff_context * 10) + (ch - '0');
105 			break;
106 		case 'a':
107 			dflags |= D_FORCEASCII;
108 			break;
109 		case 'b':
110 			dflags |= D_FOLDBLANKS;
111 			break;
112 		case 'C':
113 		case 'c':
114 			diff_format = D_CONTEXT;
115 			if (optarg != NULL) {
116 				l = strtol(optarg, &ep, 10);
117 				if (*ep != '\0' || l < 0 || l >= INT_MAX)
118 					usage();
119 				diff_context = (int)l;
120 			} else
121 				diff_context = 3;
122 			break;
123 		case 'd':
124 			dflags |= D_MINIMAL;
125 			break;
126 		case 'D':
127 			diff_format = D_IFDEF;
128 			ifdefname = optarg;
129 			break;
130 		case 'e':
131 			diff_format = D_EDIT;
132 			break;
133 		case 'f':
134 			diff_format = D_REVERSE;
135 			break;
136 		case 'h':
137 			/* silently ignore for backwards compatibility */
138 			break;
139 		case 'I':
140 			push_ignore_pats(optarg);
141 			break;
142 		case 'i':
143 			dflags |= D_IGNORECASE;
144 			break;
145 		case 'L':
146 			if (label[0] == NULL)
147 				label[0] = optarg;
148 			else if (label[1] == NULL)
149 				label[1] = optarg;
150 			else
151 				usage();
152 			break;
153 		case 'N':
154 			Nflag = 1;
155 			break;
156 		case 'n':
157 			diff_format = D_NREVERSE;
158 			break;
159 		case 'p':
160 			dflags |= D_PROTOTYPE;
161 			break;
162 		case 'P':
163 			Pflag = 1;
164 			break;
165 		case 'r':
166 			rflag = 1;
167 			break;
168 		case 'q':
169 			diff_format = D_BRIEF;
170 			break;
171 		case 'S':
172 			start = optarg;
173 			break;
174 		case 's':
175 			sflag = 1;
176 			break;
177 		case 'T':
178 			Tflag = 1;
179 			break;
180 		case 't':
181 			dflags |= D_EXPANDTABS;
182 			break;
183 		case 'U':
184 		case 'u':
185 			diff_format = D_UNIFIED;
186 			if (optarg != NULL) {
187 				l = strtol(optarg, &ep, 10);
188 				if (*ep != '\0' || l < 0 || l >= INT_MAX)
189 					usage();
190 				diff_context = (int)l;
191 			} else
192 				diff_context = 3;
193 			break;
194 		case 'w':
195 			dflags |= D_IGNOREBLANKS;
196 			break;
197 		case 'X':
198 			read_excludes_file(optarg);
199 			break;
200 		case 'x':
201 			push_excludes(optarg);
202 			break;
203 		default:
204 			usage();
205 			break;
206 		}
207 		lastch = ch;
208 		newarg = optind != prevoptind;
209 		prevoptind = optind;
210 	}
211 	argc -= optind;
212 	argv += optind;
213 
214 	if (pledge("stdio rpath tmppath", NULL) == -1)
215 		err(2, "pledge");
216 
217 	/*
218 	 * Do sanity checks, fill in stb1 and stb2 and call the appropriate
219 	 * driver routine.  Both drivers use the contents of stb1 and stb2.
220 	 */
221 	if (argc != 2)
222 		usage();
223 	if (ignore_pats != NULL) {
224 		char buf[BUFSIZ];
225 		int error;
226 
227 		if ((error = regcomp(&ignore_re, ignore_pats,
228 				     REG_NEWLINE | REG_EXTENDED)) != 0) {
229 			regerror(error, &ignore_re, buf, sizeof(buf));
230 			if (*ignore_pats != '\0')
231 				errx(2, "%s: %s", ignore_pats, buf);
232 			else
233 				errx(2, "%s", buf);
234 		}
235 	}
236 	if (strcmp(argv[0], "-") == 0) {
237 		fstat(STDIN_FILENO, &stb1);
238 		gotstdin = 1;
239 	} else if (stat(argv[0], &stb1) != 0)
240 		err(2, "%s", argv[0]);
241 	if (strcmp(argv[1], "-") == 0) {
242 		fstat(STDIN_FILENO, &stb2);
243 		gotstdin = 1;
244 	} else if (stat(argv[1], &stb2) != 0)
245 		err(2, "%s", argv[1]);
246 	if (gotstdin && (S_ISDIR(stb1.st_mode) || S_ISDIR(stb2.st_mode)))
247 		errx(2, "can't compare - to a directory");
248 	set_argstr(oargv, argv);
249 	if (S_ISDIR(stb1.st_mode) && S_ISDIR(stb2.st_mode)) {
250 		if (diff_format == D_IFDEF)
251 			errx(2, "-D option not supported with directories");
252 		diffdir(argv[0], argv[1], dflags);
253 	} else {
254 		if (S_ISDIR(stb1.st_mode)) {
255 			argv[0] = splice(argv[0], argv[1]);
256 			if (stat(argv[0], &stb1) == -1)
257 				err(2, "%s", argv[0]);
258 		}
259 		if (S_ISDIR(stb2.st_mode)) {
260 			argv[1] = splice(argv[1], argv[0]);
261 			if (stat(argv[1], &stb2) == -1)
262 				err(2, "%s", argv[1]);
263 		}
264 		print_status(diffreg(argv[0], argv[1], dflags), argv[0], argv[1],
265 		    "");
266 	}
267 	exit(status);
268 }
269 
270 void
271 set_argstr(char **av, char **ave)
272 {
273 	size_t argsize;
274 	char **ap;
275 
276 	argsize = 4 + *ave - *av + 1;
277 	diffargs = xmalloc(argsize);
278 	strlcpy(diffargs, "diff", argsize);
279 	for (ap = av + 1; ap < ave; ap++) {
280 		if (strcmp(*ap, "--") != 0) {
281 			strlcat(diffargs, " ", argsize);
282 			strlcat(diffargs, *ap, argsize);
283 		}
284 	}
285 }
286 
287 /*
288  * Read in an excludes file and push each line.
289  */
290 void
291 read_excludes_file(char *file)
292 {
293 	FILE *fp;
294 	char *buf, *pattern;
295 	size_t len;
296 
297 	if (strcmp(file, "-") == 0)
298 		fp = stdin;
299 	else if ((fp = fopen(file, "r")) == NULL)
300 		err(2, "%s", file);
301 	while ((buf = fgetln(fp, &len)) != NULL) {
302 		if (buf[len - 1] == '\n')
303 			len--;
304 		pattern = xmalloc(len + 1);
305 		memcpy(pattern, buf, len);
306 		pattern[len] = '\0';
307 		push_excludes(pattern);
308 	}
309 	if (strcmp(file, "-") != 0)
310 		fclose(fp);
311 }
312 
313 /*
314  * Push a pattern onto the excludes list.
315  */
316 void
317 push_excludes(char *pattern)
318 {
319 	struct excludes *entry;
320 
321 	entry = xmalloc(sizeof(*entry));
322 	entry->pattern = pattern;
323 	entry->next = excludes_list;
324 	excludes_list = entry;
325 }
326 
327 void
328 push_ignore_pats(char *pattern)
329 {
330 	size_t len;
331 
332 	if (ignore_pats == NULL)
333 		ignore_pats = xstrdup(pattern);
334 	else {
335 		/* old + "|" + new + NUL */
336 		len = strlen(ignore_pats) + strlen(pattern) + 2;
337 		ignore_pats = xreallocarray(ignore_pats, 1, len);
338 		strlcat(ignore_pats, "|", len);
339 		strlcat(ignore_pats, pattern, len);
340 	}
341 }
342 
343 void
344 print_only(const char *path, size_t dirlen, const char *entry)
345 {
346 	if (dirlen > 1)
347 		dirlen--;
348 	printf("Only in %.*s: %s\n", (int)dirlen, path, entry);
349 }
350 
351 void
352 print_status(int val, char *path1, char *path2, char *entry)
353 {
354 	switch (val) {
355 	case D_BINARY:
356 		printf("Binary files %s%s and %s%s differ\n",
357 		    path1, entry, path2, entry);
358 		break;
359 	case D_DIFFER:
360 		if (diff_format == D_BRIEF)
361 			printf("Files %s%s and %s%s differ\n",
362 			    path1, entry, path2, entry);
363 		break;
364 	case D_SAME:
365 		if (sflag)
366 			printf("Files %s%s and %s%s are identical\n",
367 			    path1, entry, path2, entry);
368 		break;
369 	case D_MISMATCH1:
370 		printf("File %s%s is a directory while file %s%s is a regular file\n",
371 		    path1, entry, path2, entry);
372 		break;
373 	case D_MISMATCH2:
374 		printf("File %s%s is a regular file while file %s%s is a directory\n",
375 		    path1, entry, path2, entry);
376 		break;
377 	case D_SKIPPED1:
378 		printf("File %s%s is not a regular file or directory and was skipped\n",
379 		    path1, entry);
380 		break;
381 	case D_SKIPPED2:
382 		printf("File %s%s is not a regular file or directory and was skipped\n",
383 		    path2, entry);
384 		break;
385 	}
386 }
387 
388 __dead void
389 usage(void)
390 {
391 	(void)fprintf(stderr,
392 	    "usage: diff [-abdipTtw] [-c | -e | -f | -n | -q | -u] [-I pattern] [-L label]\n"
393 	    "            file1 file2\n"
394 	    "       diff [-abdipTtw] [-I pattern] [-L label] -C number file1 file2\n"
395 	    "       diff [-abditw] [-I pattern] -D string file1 file2\n"
396 	    "       diff [-abdipTtw] [-I pattern] [-L label] -U number file1 file2\n"
397 	    "       diff [-abdiNPprsTtw] [-c | -e | -f | -n | -q | -u] [-I pattern]\n"
398 	    "            [-L label] [-S name] [-X file] [-x pattern] dir1 dir2\n");
399 
400 	exit(2);
401 }
402