xref: /openbsd/usr.bin/patch/patch.c (revision d756b25f)
1 /*	$OpenBSD: patch.c,v 1.76 2024/03/22 19:22:23 jcs Exp $	*/
2 
3 /*
4  * patch - a program to apply diffs to original files
5  *
6  * Copyright 1986, Larry Wall
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following condition is met:
10  * 1. Redistributions of source code must retain the above copyright notice,
11  * this condition and the following disclaimer.
12  *
13  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND ANY
14  * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
15  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
16  * DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
17  * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
18  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
19  * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
20  * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
21  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
22  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
23  * SUCH DAMAGE.
24  *
25  * -C option added in 1998, original code by Marc Espie, based on FreeBSD
26  * behaviour
27  */
28 
29 #include <sys/types.h>
30 #include <sys/stat.h>
31 #include <unistd.h>
32 
33 #include <ctype.h>
34 #include <getopt.h>
35 #include <libgen.h>
36 #include <limits.h>
37 #include <paths.h>
38 #include <stdio.h>
39 #include <string.h>
40 #include <stdlib.h>
41 
42 #include "common.h"
43 #include "util.h"
44 #include "pch.h"
45 #include "inp.h"
46 #include "backupfile.h"
47 #include "ed.h"
48 
49 mode_t		filemode = 0644;
50 
51 char		*buf;			/* general purpose buffer */
52 size_t		 bufsz;			/* general purpose buffer size */
53 
54 bool		using_plan_a = true;	/* try to keep everything in memory */
55 bool		out_of_mem = false;	/* ran out of memory in plan a */
56 
57 #define MAXFILEC 2
58 
59 char		*filearg[MAXFILEC];
60 bool		ok_to_create_file = false;
61 char		*outname = NULL;
62 char		*origprae = NULL;
63 char		*TMPOUTNAME;
64 char		*TMPINNAME;
65 char		*TMPREJNAME;
66 char		*TMPPATNAME;
67 bool		toutkeep = false;
68 bool		trejkeep = false;
69 bool		warn_on_invalid_line;
70 bool		last_line_missing_eol;
71 
72 #ifdef DEBUGGING
73 int		debug = 0;
74 #endif
75 
76 bool		force = false;
77 bool		batch = false;
78 bool		verbose = true;
79 bool		reverse = false;
80 bool		noreverse = false;
81 bool		skip_rest_of_patch = false;
82 int		strippath = 957;
83 bool		canonicalize = false;
84 bool		check_only = false;
85 int		diff_type = 0;
86 char		*revision = NULL;	/* prerequisite revision, if any */
87 LINENUM		input_lines = 0;	/* how long is input file in lines */
88 int		posix = 0;		/* strict POSIX mode? */
89 
90 static void	reinitialize_almost_everything(void);
91 static void	get_some_switches(void);
92 static LINENUM	locate_hunk(LINENUM);
93 static void	abort_context_hunk(void);
94 static void	rej_line(int, LINENUM);
95 static void	abort_hunk(void);
96 static void	apply_hunk(LINENUM);
97 static void	init_output(const char *);
98 static void	init_reject(const char *);
99 static void	copy_till(LINENUM, bool);
100 static void	spew_output(void);
101 static void	dump_line(LINENUM, bool);
102 static bool	patch_match(LINENUM, LINENUM, LINENUM);
103 static bool	similar(const char *, const char *, ssize_t);
104 static __dead void usage(void);
105 
106 /* true if -E was specified on command line.  */
107 static bool	remove_empty_files = false;
108 
109 /* true if -R was specified on command line.  */
110 static bool	reverse_flag_specified = false;
111 
112 static bool	Vflag = false;
113 
114 /* buffer holding the name of the rejected patch file. */
115 static char	rejname[PATH_MAX];
116 
117 /* how many input lines have been irretractibly output */
118 static LINENUM	last_frozen_line = 0;
119 
120 static int	Argc;		/* guess */
121 static char	**Argv;
122 static int	Argc_last;	/* for restarting plan_b */
123 static char	**Argv_last;
124 
125 static FILE	*ofp = NULL;	/* output file pointer */
126 static FILE	*rejfp = NULL;	/* reject file pointer */
127 
128 static int	filec = 0;	/* how many file arguments? */
129 static LINENUM	last_offset = 0;
130 static LINENUM	maxfuzz = 2;
131 
132 /* patch using ifdef, ifndef, etc. */
133 static bool		do_defines = false;
134 /* #ifdef xyzzy */
135 static char		if_defined[128];
136 /* #ifndef xyzzy */
137 static char		not_defined[128];
138 /* #else */
139 static const char	else_defined[] = "#else\n";
140 /* #endif xyzzy */
141 static char		end_defined[128];
142 
143 
144 /* Apply a set of diffs as appropriate. */
145 
146 int
main(int argc,char * argv[])147 main(int argc, char *argv[])
148 {
149 	int	error = 0, hunk, failed, i, fd;
150 	bool	patch_seen;
151 	LINENUM	where = 0, newwhere, fuzz, mymaxfuzz;
152 	const	char *tmpdir;
153 	char	*v;
154 
155 	if (pledge("stdio rpath wpath cpath tmppath fattr unveil", NULL) == -1) {
156 		perror("pledge");
157 		my_exit(2);
158 	}
159 
160 	bufsz = INITLINELEN;
161 	if ((buf = malloc(bufsz)) == NULL)
162 		pfatal("allocating input buffer");
163 	buf[0] = '\0';
164 
165 	setvbuf(stdout, NULL, _IOLBF, 0);
166 	setvbuf(stderr, NULL, _IOLBF, 0);
167 	for (i = 0; i < MAXFILEC; i++)
168 		filearg[i] = NULL;
169 
170 	/* Cons up the names of the temporary files.  */
171 	if ((tmpdir = getenv("TMPDIR")) == NULL || *tmpdir == '\0')
172 		tmpdir = _PATH_TMP;
173 	for (i = strlen(tmpdir) - 1; i > 0 && tmpdir[i] == '/'; i--)
174 		;
175 	i++;
176 	if (asprintf(&TMPOUTNAME, "%.*s/patchoXXXXXXXXXX", i, tmpdir) == -1)
177 		fatal("cannot allocate memory");
178 	if ((fd = mkstemp(TMPOUTNAME)) == -1)
179 		pfatal("can't create %s", TMPOUTNAME);
180 	close(fd);
181 
182 	if (asprintf(&TMPINNAME, "%.*s/patchiXXXXXXXXXX", i, tmpdir) == -1)
183 		fatal("cannot allocate memory");
184 	if ((fd = mkstemp(TMPINNAME)) == -1)
185 		pfatal("can't create %s", TMPINNAME);
186 	close(fd);
187 
188 	if (asprintf(&TMPREJNAME, "%.*s/patchrXXXXXXXXXX", i, tmpdir) == -1)
189 		fatal("cannot allocate memory");
190 	if ((fd = mkstemp(TMPREJNAME)) == -1)
191 		pfatal("can't create %s", TMPREJNAME);
192 	close(fd);
193 
194 	if (asprintf(&TMPPATNAME, "%.*s/patchpXXXXXXXXXX", i, tmpdir) == -1)
195 		fatal("cannot allocate memory");
196 	if ((fd = mkstemp(TMPPATNAME)) == -1)
197 		pfatal("can't create %s", TMPPATNAME);
198 	close(fd);
199 
200 	v = getenv("SIMPLE_BACKUP_SUFFIX");
201 	if (v)
202 		simple_backup_suffix = v;
203 	else
204 		simple_backup_suffix = ORIGEXT;
205 
206 	/* parse switches */
207 	Argc = argc;
208 	Argv = argv;
209 	get_some_switches();
210 	if (unveil(tmpdir, "rwc") == -1) {
211 		perror("unveil");
212 		my_exit(2);
213 	}
214 	if (outname != NULL)
215 		if (unveil(outname, "rwc") == -1) {
216 			perror("unveil");
217 			my_exit(2);
218 		}
219 	if (filearg[0] != NULL) {
220 		char *origdir;
221 
222 		if (unveil(filearg[0], "rwc") == -1) {
223 			perror("unveil");
224 			my_exit(2);
225 		}
226 		if ((origdir = dirname(filearg[0])) == NULL) {
227 			perror("dirname");
228 			my_exit(2);
229 		}
230 		if (unveil(origdir, "rwc") == -1) {
231 			perror("unveil");
232 			my_exit(2);
233 		}
234 	} else {
235 		if (unveil(".", "rwc") == -1) {
236 			perror("unveil");
237 			my_exit(2);
238 		}
239 	}
240 	if (filearg[1] != NULL)
241 		if (unveil(filearg[1], "r") == -1) {
242 			perror("unveil");
243 			my_exit(2);
244 		}
245 	if (!force && !batch)
246 		if (unveil(_PATH_TTY, "r") == -1) {
247 			perror("unveil");
248 			my_exit(2);
249 		}
250 	if (*rejname != '\0')
251 		if (unveil(rejname, "rwc") == -1) {
252 			perror("unveil");
253 			my_exit(2);
254 		}
255 	if (pledge("stdio rpath wpath cpath tmppath fattr", NULL) == -1) {
256 		perror("pledge");
257 		my_exit(2);
258 	}
259 
260 	if (!Vflag) {
261 		if ((v = getenv("PATCH_VERSION_CONTROL")) == NULL)
262 			v = getenv("VERSION_CONTROL");
263 		if (v != NULL || !posix)
264 			backup_type = get_version(v);	/* OK to pass NULL. */
265 	}
266 
267 	/* make sure we clean up /tmp in case of disaster */
268 	set_signals(0);
269 
270 	patch_seen = false;
271 	for (open_patch_file(filearg[1]); there_is_another_patch();
272 	    reinitialize_almost_everything()) {
273 		/* for each patch in patch file */
274 
275 		patch_seen = true;
276 
277 		warn_on_invalid_line = true;
278 
279 		if (outname == NULL)
280 			outname = xstrdup(filearg[0]);
281 
282 		/* initialize the patched file */
283 		if (!skip_rest_of_patch)
284 			init_output(TMPOUTNAME);
285 
286 		/* initialize reject file */
287 		init_reject(TMPREJNAME);
288 
289 		/* find out where all the lines are */
290 		if (!skip_rest_of_patch)
291 			scan_input(filearg[0]);
292 
293 		/* for ed script just up and do it and exit */
294 		if (diff_type == ED_DIFF) {
295 			do_ed_script();
296 			continue;
297 		}
298 
299 		/* from here on, open no standard i/o files, because malloc */
300 		/* might misfire and we can't catch it easily */
301 
302 		/* apply each hunk of patch */
303 		hunk = 0;
304 		failed = 0;
305 		out_of_mem = false;
306 		while (another_hunk()) {
307 			hunk++;
308 			fuzz = 0;
309 			mymaxfuzz = pch_context();
310 			if (maxfuzz < mymaxfuzz)
311 				mymaxfuzz = maxfuzz;
312 			if (!skip_rest_of_patch) {
313 				do {
314 					where = locate_hunk(fuzz);
315 					if ((hunk == 1 && where == 0 && !force) ||
316 					    (where == 1 && pch_ptrn_lines() == 0 && !force)) {
317 						/* dwim for reversed patch? */
318 						if (!pch_swap()) {
319 							if (fuzz == 0)
320 								say("Not enough memory to try swapped hunk!  Assuming unswapped.\n");
321 							continue;
322 						}
323 						reverse = !reverse;
324 						/* try again */
325 						where = locate_hunk(fuzz);
326 						if (where == 0) {
327 							/* didn't find it swapped */
328 							if (!pch_swap())
329 								/* put it back to normal */
330 								fatal("lost hunk on alloc error!\n");
331 							reverse = !reverse;
332 
333 							/* restore position if this patch creates a file */
334 							if (pch_ptrn_lines() == 0)
335 								where = 1;
336 						} else if (noreverse) {
337 							if (!pch_swap())
338 								/* put it back to normal */
339 								fatal("lost hunk on alloc error!\n");
340 							reverse = !reverse;
341 							say("Ignoring previously applied (or reversed) patch.\n");
342 							skip_rest_of_patch = true;
343 						} else if (batch) {
344 							if (verbose)
345 								say("%seversed (or previously applied) patch detected!  %s -R.",
346 								    reverse ? "R" : "Unr",
347 								    reverse ? "Assuming" : "Ignoring");
348 						} else {
349 							ask("%seversed (or previously applied) patch detected!  %s -R? [y] ",
350 							    reverse ? "R" : "Unr",
351 							    reverse ? "Assume" : "Ignore");
352 							if (*buf == 'n') {
353 								ask("Apply anyway? [n] ");
354 								if (*buf != 'y')
355 									skip_rest_of_patch = true;
356 								where = 0;
357 								reverse = !reverse;
358 								if (!pch_swap())
359 									/* put it back to normal */
360 									fatal("lost hunk on alloc error!\n");
361 							}
362 						}
363 					}
364 				} while (!skip_rest_of_patch && where == 0 &&
365 				    ++fuzz <= mymaxfuzz);
366 
367 				if (skip_rest_of_patch) {	/* just got decided */
368 					fclose(ofp);
369 					ofp = NULL;
370 				}
371 			}
372 			newwhere = pch_newfirst() + last_offset;
373 			if (skip_rest_of_patch) {
374 				abort_hunk();
375 				failed++;
376 				if (verbose)
377 					say("Hunk #%d ignored at %ld.\n",
378 					    hunk, newwhere);
379 			} else if (where == 0) {
380 				abort_hunk();
381 				failed++;
382 				if (verbose)
383 					say("Hunk #%d failed at %ld.\n",
384 					    hunk, newwhere);
385 			} else {
386 				apply_hunk(where);
387 				if (verbose) {
388 					say("Hunk #%d succeeded at %ld",
389 					    hunk, newwhere);
390 					if (fuzz != 0)
391 						say(" with fuzz %ld", fuzz);
392 					if (last_offset)
393 						say(" (offset %ld line%s)",
394 						    last_offset,
395 						    last_offset == 1L ? "" : "s");
396 					say(".\n");
397 				}
398 			}
399 		}
400 
401 		if (out_of_mem && using_plan_a) {
402 			Argc = Argc_last;
403 			Argv = Argv_last;
404 			say("\n\nRan out of memory using Plan A--trying again...\n\n");
405 			if (ofp)
406 				fclose(ofp);
407 			ofp = NULL;
408 			if (rejfp)
409 				fclose(rejfp);
410 			rejfp = NULL;
411 			continue;
412 		}
413 		if (hunk == 0)
414 			fatal("Internal error: hunk should not be 0\n");
415 
416 		/* finish spewing out the new file */
417 		if (!skip_rest_of_patch)
418 			spew_output();
419 
420 		/* and put the output where desired */
421 		ignore_signals();
422 		if (!skip_rest_of_patch) {
423 			struct stat	statbuf;
424 			char	*realout = outname;
425 
426 			if (!check_only) {
427 				if (move_file(TMPOUTNAME, outname) < 0) {
428 					toutkeep = true;
429 					realout = TMPOUTNAME;
430 					chmod(TMPOUTNAME, filemode);
431 				} else
432 					chmod(outname, filemode);
433 
434 				if (remove_empty_files &&
435 				    stat(realout, &statbuf) == 0 &&
436 				    statbuf.st_size == 0) {
437 					if (verbose)
438 						say("Removing %s (empty after patching).\n",
439 						    realout);
440 					unlink(realout);
441 				}
442 			}
443 		}
444 		fclose(rejfp);
445 		rejfp = NULL;
446 		if (failed) {
447 			error = 1;
448 			if (*rejname == '\0') {
449 				if (strlcpy(rejname, outname,
450 				    sizeof(rejname)) >= sizeof(rejname))
451 					fatal("filename %s is too long\n", outname);
452 				if (strlcat(rejname, REJEXT,
453 				    sizeof(rejname)) >= sizeof(rejname))
454 					fatal("filename %s is too long\n", outname);
455 			}
456 			if (!check_only)
457 				say("%d out of %d hunks %s--saving rejects to %s\n",
458 				    failed, hunk, skip_rest_of_patch ? "ignored" : "failed", rejname);
459 			else
460 				say("%d out of %d hunks %s\n",
461 				    failed, hunk, skip_rest_of_patch ? "ignored" : "failed");
462 			if (!check_only && move_file(TMPREJNAME, rejname) < 0)
463 				trejkeep = true;
464 		}
465 		set_signals(1);
466 	}
467 
468 	if (!patch_seen)
469 		error = 2;
470 
471 	my_exit(error);
472 	/* NOTREACHED */
473 }
474 
475 /* Prepare to find the next patch to do in the patch file. */
476 
477 static void
reinitialize_almost_everything(void)478 reinitialize_almost_everything(void)
479 {
480 	re_patch();
481 	re_input();
482 
483 	input_lines = 0;
484 	last_frozen_line = 0;
485 
486 	filec = 0;
487 	if (!out_of_mem) {
488 		free(filearg[0]);
489 		filearg[0] = NULL;
490 	}
491 
492 	free(outname);
493 	outname = NULL;
494 
495 	last_offset = 0;
496 	diff_type = 0;
497 
498 	free(revision);
499 	revision = NULL;
500 
501 	reverse = reverse_flag_specified;
502 	skip_rest_of_patch = false;
503 
504 	get_some_switches();
505 }
506 
507 /* Process switches and filenames. */
508 
509 static void
get_some_switches(void)510 get_some_switches(void)
511 {
512 	const char *options = "b::B:cCd:D:eEfF:i:lnNo:p:r:RstuvV:x:z:";
513 	static struct option longopts[] = {
514 		{"backup",		no_argument,		0,	'b'},
515 		{"batch",		no_argument,		0,	't'},
516 		{"check",		no_argument,		0,	'C'},
517 		{"context",		no_argument,		0,	'c'},
518 		{"debug",		required_argument,	0,	'x'},
519 		{"directory",		required_argument,	0,	'd'},
520 		{"dry-run",		no_argument,		0,	'C'},
521 		{"ed",			no_argument,		0,	'e'},
522 		{"force",		no_argument,		0,	'f'},
523 		{"forward",		no_argument,		0,	'N'},
524 		{"fuzz",		required_argument,	0,	'F'},
525 		{"ifdef",		required_argument,	0,	'D'},
526 		{"input",		required_argument,	0,	'i'},
527 		{"ignore-whitespace",	no_argument,		0,	'l'},
528 		{"normal",		no_argument,		0,	'n'},
529 		{"output",		required_argument,	0,	'o'},
530 		{"prefix",		required_argument,	0,	'B'},
531 		{"quiet",		no_argument,		0,	's'},
532 		{"reject-file",		required_argument,	0,	'r'},
533 		{"remove-empty-files",	no_argument,		0,	'E'},
534 		{"reverse",		no_argument,		0,	'R'},
535 		{"silent",		no_argument,		0,	's'},
536 		{"strip",		required_argument,	0,	'p'},
537 		{"suffix",		required_argument,	0,	'z'},
538 		{"unified",		no_argument,		0,	'u'},
539 		{"version",		no_argument,		0,	'v'},
540 		{"version-control",	required_argument,	0,	'V'},
541 		{"posix",		no_argument,		&posix,	1},
542 		{NULL,			0,			0,	0}
543 	};
544 	int ch;
545 
546 	rejname[0] = '\0';
547 	Argc_last = Argc;
548 	Argv_last = Argv;
549 	if (!Argc)
550 		return;
551 	optreset = optind = 1;
552 	while ((ch = getopt_long(Argc, Argv, options, longopts, NULL)) != -1) {
553 		switch (ch) {
554 		case 'b':
555 			if (backup_type == none)
556 				backup_type = numbered_existing;
557 			if (optarg == NULL)
558 				break;
559 			if (verbose)
560 				say("Warning, the ``-b suffix'' option has been"
561 				    " obsoleted by the -z option.\n");
562 			/* FALLTHROUGH */
563 		case 'z':
564 			/* must directly follow 'b' case for backwards compat */
565 			simple_backup_suffix = xstrdup(optarg);
566 			break;
567 		case 'B':
568 			origprae = xstrdup(optarg);
569 			break;
570 		case 'c':
571 			diff_type = CONTEXT_DIFF;
572 			break;
573 		case 'C':
574 			check_only = true;
575 			break;
576 		case 'd':
577 			if (chdir(optarg) == -1)
578 				pfatal("can't cd to %s", optarg);
579 			break;
580 		case 'D':
581 			do_defines = true;
582 			if (!isalpha((unsigned char)*optarg) && *optarg != '_')
583 				fatal("argument to -D is not an identifier\n");
584 			snprintf(if_defined, sizeof if_defined,
585 			    "#ifdef %s\n", optarg);
586 			snprintf(not_defined, sizeof not_defined,
587 			    "#ifndef %s\n", optarg);
588 			snprintf(end_defined, sizeof end_defined,
589 			    "#endif /* %s */\n", optarg);
590 			break;
591 		case 'e':
592 			diff_type = ED_DIFF;
593 			break;
594 		case 'E':
595 			remove_empty_files = true;
596 			break;
597 		case 'f':
598 			force = true;
599 			break;
600 		case 'F':
601 			maxfuzz = atoi(optarg);
602 			break;
603 		case 'i':
604 			if (++filec == MAXFILEC)
605 				fatal("too many file arguments\n");
606 			filearg[filec] = xstrdup(optarg);
607 			break;
608 		case 'l':
609 			canonicalize = true;
610 			break;
611 		case 'n':
612 			diff_type = NORMAL_DIFF;
613 			break;
614 		case 'N':
615 			noreverse = true;
616 			break;
617 		case 'o':
618 			outname = xstrdup(optarg);
619 			break;
620 		case 'p':
621 			strippath = atoi(optarg);
622 			break;
623 		case 'r':
624 			if (strlcpy(rejname, optarg,
625 			    sizeof(rejname)) >= sizeof(rejname))
626 				fatal("argument for -r is too long\n");
627 			break;
628 		case 'R':
629 			reverse = true;
630 			reverse_flag_specified = true;
631 			break;
632 		case 's':
633 			verbose = false;
634 			break;
635 		case 't':
636 			batch = true;
637 			break;
638 		case 'u':
639 			diff_type = UNI_DIFF;
640 			break;
641 		case 'v':
642 			version();
643 			break;
644 		case 'V':
645 			backup_type = get_version(optarg);
646 			Vflag = true;
647 			break;
648 #ifdef DEBUGGING
649 		case 'x':
650 			debug = atoi(optarg);
651 			break;
652 #endif
653 		default:
654 			if (ch != '\0')
655 				usage();
656 			break;
657 		}
658 	}
659 	Argc -= optind;
660 	Argv += optind;
661 
662 	if (Argc > 0) {
663 		filearg[0] = xstrdup(*Argv++);
664 		Argc--;
665 		while (Argc > 0) {
666 			if (++filec == MAXFILEC)
667 				fatal("too many file arguments\n");
668 			filearg[filec] = xstrdup(*Argv++);
669 			Argc--;
670 		}
671 	}
672 
673 	if (getenv("POSIXLY_CORRECT") != NULL)
674 		posix = 1;
675 }
676 
677 static __dead void
usage(void)678 usage(void)
679 {
680 	fprintf(stderr,
681 "usage: patch [-bCcEeflNnRstuv] [-B backup-prefix] [-D symbol] [-d directory]\n"
682 "             [-F max-fuzz] [-i patchfile] [-o out-file] [-p strip-count]\n"
683 "             [-r rej-name] [-V t | nil | never | none] [-x number]\n"
684 "             [-z backup-ext] [--posix] [origfile [patchfile]]\n"
685 "       patch <patchfile\n");
686 	my_exit(2);
687 }
688 
689 /*
690  * Attempt to find the right place to apply this hunk of patch.
691  */
692 static LINENUM
locate_hunk(LINENUM fuzz)693 locate_hunk(LINENUM fuzz)
694 {
695 	LINENUM	first_guess = pch_first() + last_offset;
696 	LINENUM	offset;
697 	LINENUM	pat_lines = pch_ptrn_lines();
698 	LINENUM	max_pos_offset = input_lines - first_guess - pat_lines + 1;
699 	LINENUM	max_neg_offset = first_guess - last_frozen_line - 1 + pch_context();
700 
701 	if (pat_lines == 0) {		/* null range matches always */
702 		if (verbose && fuzz == 0 && (diff_type == CONTEXT_DIFF
703 		    || diff_type == NEW_CONTEXT_DIFF
704 		    || diff_type == UNI_DIFF)) {
705 			say("Empty context always matches.\n");
706 		}
707 		if (first_guess == 0)
708 			return 1;
709 		return (first_guess);
710 	}
711 	if (max_neg_offset >= first_guess)	/* do not try lines < 0 */
712 		max_neg_offset = first_guess - 1;
713 	if (first_guess <= input_lines && patch_match(first_guess, 0, fuzz))
714 		return first_guess;
715 	for (offset = 1; ; offset++) {
716 		bool	check_after = (offset <= max_pos_offset);
717 		bool	check_before = (offset <= max_neg_offset);
718 
719 		if (check_after && patch_match(first_guess, offset, fuzz)) {
720 #ifdef DEBUGGING
721 			if (debug & 1)
722 				say("Offset changing from %ld to %ld\n",
723 				    last_offset, offset);
724 #endif
725 			last_offset = offset;
726 			return first_guess + offset;
727 		} else if (check_before && patch_match(first_guess, -offset, fuzz)) {
728 #ifdef DEBUGGING
729 			if (debug & 1)
730 				say("Offset changing from %ld to %ld\n",
731 				    last_offset, -offset);
732 #endif
733 			last_offset = -offset;
734 			return first_guess - offset;
735 		} else if (!check_before && !check_after)
736 			return 0;
737 	}
738 }
739 
740 /* We did not find the pattern, dump out the hunk so they can handle it. */
741 
742 static void
abort_context_hunk(void)743 abort_context_hunk(void)
744 {
745 	LINENUM	i;
746 	const LINENUM	pat_end = pch_end();
747 	/*
748 	 * add in last_offset to guess the same as the previous successful
749 	 * hunk
750 	 */
751 	const LINENUM	oldfirst = pch_first() + last_offset;
752 	const LINENUM	newfirst = pch_newfirst() + last_offset;
753 	const LINENUM	oldlast = oldfirst + pch_ptrn_lines() - 1;
754 	const LINENUM	newlast = newfirst + pch_repl_lines() - 1;
755 	const char	*stars = (diff_type >= NEW_CONTEXT_DIFF ? " ****" : "");
756 	const char	*minuses = (diff_type >= NEW_CONTEXT_DIFF ? " ----" : " -----");
757 
758 	fprintf(rejfp, "***************\n");
759 	for (i = 0; i <= pat_end; i++) {
760 		switch (pch_char(i)) {
761 		case '*':
762 			if (oldlast < oldfirst)
763 				fprintf(rejfp, "*** 0%s\n", stars);
764 			else if (oldlast == oldfirst)
765 				fprintf(rejfp, "*** %ld%s\n", oldfirst, stars);
766 			else
767 				fprintf(rejfp, "*** %ld,%ld%s\n", oldfirst,
768 				    oldlast, stars);
769 			break;
770 		case '=':
771 			if (newlast < newfirst)
772 				fprintf(rejfp, "--- 0%s\n", minuses);
773 			else if (newlast == newfirst)
774 				fprintf(rejfp, "--- %ld%s\n", newfirst, minuses);
775 			else
776 				fprintf(rejfp, "--- %ld,%ld%s\n", newfirst,
777 				    newlast, minuses);
778 			break;
779 		case '\n':
780 			fprintf(rejfp, "%s", pfetch(i));
781 			break;
782 		case ' ':
783 		case '-':
784 		case '+':
785 		case '!':
786 			fprintf(rejfp, "%c %s", pch_char(i), pfetch(i));
787 			break;
788 		default:
789 			fatal("fatal internal error in abort_context_hunk\n");
790 		}
791 	}
792 }
793 
794 static void
rej_line(int ch,LINENUM i)795 rej_line(int ch, LINENUM i)
796 {
797 	size_t len;
798 	const char *line = pfetch(i);
799 
800 	len = strlen(line);
801 
802 	fprintf(rejfp, "%c%s", ch, line);
803 	if (len == 0 || line[len-1] != '\n')
804 		fprintf(rejfp, "\n\\ No newline at end of file\n");
805 }
806 
807 static void
abort_hunk(void)808 abort_hunk(void)
809 {
810 	LINENUM		i, j, split;
811 	int		ch1, ch2;
812 	const LINENUM	pat_end = pch_end();
813 	const LINENUM	oldfirst = pch_first() + last_offset;
814 	const LINENUM	newfirst = pch_newfirst() + last_offset;
815 
816 	if (diff_type != UNI_DIFF) {
817 		abort_context_hunk();
818 		return;
819 	}
820 	split = -1;
821 	for (i = 0; i <= pat_end; i++) {
822 		if (pch_char(i) == '=') {
823 			split = i;
824 			break;
825 		}
826 	}
827 	if (split == -1) {
828 		fprintf(rejfp, "malformed hunk: no split found\n");
829 		return;
830 	}
831 	i = 0;
832 	j = split + 1;
833 	fprintf(rejfp, "@@ -%ld,%ld +%ld,%ld @@\n",
834 	    pch_ptrn_lines() ? oldfirst : 0,
835 	    pch_ptrn_lines(), newfirst, pch_repl_lines());
836 	while (i < split || j <= pat_end) {
837 		ch1 = i < split ? pch_char(i) : -1;
838 		ch2 = j <= pat_end ? pch_char(j) : -1;
839 		if (ch1 == '-') {
840 			rej_line('-', i);
841 			i++;
842 		} else if (ch1 == ' ' && ch2 == ' ') {
843 			rej_line(' ', i);
844 			i++;
845 			j++;
846 		} else if (ch1 == '!' && ch2 == '!') {
847 			while (i < split && ch1 == '!') {
848 				rej_line('-', i);
849 				i++;
850 				ch1 = i < split ? pch_char(i) : -1;
851 			}
852 			while (j <= pat_end && ch2 == '!') {
853 				rej_line('+', j);
854 				j++;
855 				ch2 = j <= pat_end ? pch_char(j) : -1;
856 			}
857 		} else if (ch1 == '*') {
858 			i++;
859 		} else if (ch2 == '+' || ch2 == ' ') {
860 			rej_line(ch2, j);
861 			j++;
862 		} else {
863 			fprintf(rejfp, "internal error on (%ld %ld %ld)\n",
864 			    i, split, j);
865 			rej_line(ch1, i);
866 			rej_line(ch2, j);
867 			return;
868 		}
869 	}
870 }
871 
872 /* We found where to apply it (we hope), so do it. */
873 
874 static void
apply_hunk(LINENUM where)875 apply_hunk(LINENUM where)
876 {
877 	LINENUM		old = 1;
878 	const LINENUM	lastline = pch_ptrn_lines();
879 	LINENUM		new = lastline + 1;
880 #define OUTSIDE 0
881 #define IN_IFNDEF 1
882 #define IN_IFDEF 2
883 #define IN_ELSE 3
884 	int		def_state = OUTSIDE;
885 	const LINENUM	pat_end = pch_end();
886 
887 	where--;
888 	while (pch_char(new) == '=' || pch_char(new) == '\n')
889 		new++;
890 
891 	while (old <= lastline) {
892 		if (pch_char(old) == '-') {
893 			copy_till(where + old - 1, false);
894 			if (do_defines) {
895 				if (def_state == OUTSIDE) {
896 					fputs(not_defined, ofp);
897 					def_state = IN_IFNDEF;
898 				} else if (def_state == IN_IFDEF) {
899 					fputs(else_defined, ofp);
900 					def_state = IN_ELSE;
901 				}
902 				fputs(pfetch(old), ofp);
903 			}
904 			last_frozen_line++;
905 			old++;
906 		} else if (new > pat_end) {
907 			break;
908 		} else if (pch_char(new) == '+') {
909 			copy_till(where + old - 1, false);
910 			if (do_defines) {
911 				if (def_state == IN_IFNDEF) {
912 					fputs(else_defined, ofp);
913 					def_state = IN_ELSE;
914 				} else if (def_state == OUTSIDE) {
915 					fputs(if_defined, ofp);
916 					def_state = IN_IFDEF;
917 				}
918 			}
919 			fputs(pfetch(new), ofp);
920 			new++;
921 		} else if (pch_char(new) != pch_char(old)) {
922 			say("Out-of-sync patch, lines %ld,%ld--mangled text or line numbers, maybe?\n",
923 			    pch_hunk_beg() + old,
924 			    pch_hunk_beg() + new);
925 #ifdef DEBUGGING
926 			say("oldchar = '%c', newchar = '%c'\n",
927 			    pch_char(old), pch_char(new));
928 #endif
929 			my_exit(2);
930 		} else if (pch_char(new) == '!') {
931 			copy_till(where + old - 1, false);
932 			if (do_defines) {
933 				fputs(not_defined, ofp);
934 				def_state = IN_IFNDEF;
935 			}
936 			while (pch_char(old) == '!') {
937 				if (do_defines) {
938 					fputs(pfetch(old), ofp);
939 				}
940 				last_frozen_line++;
941 				old++;
942 			}
943 			if (do_defines) {
944 				fputs(else_defined, ofp);
945 				def_state = IN_ELSE;
946 			}
947 			while (pch_char(new) == '!') {
948 				fputs(pfetch(new), ofp);
949 				new++;
950 			}
951 		} else {
952 			if (pch_char(new) != ' ')
953 				fatal("Internal error: expected ' '\n");
954 			old++;
955 			new++;
956 			if (do_defines && def_state != OUTSIDE) {
957 				fputs(end_defined, ofp);
958 				def_state = OUTSIDE;
959 			}
960 		}
961 	}
962 	if (new <= pat_end && pch_char(new) == '+') {
963 		copy_till(where + old - 1, false);
964 		if (do_defines) {
965 			if (def_state == OUTSIDE) {
966 				fputs(if_defined, ofp);
967 				def_state = IN_IFDEF;
968 			} else if (def_state == IN_IFNDEF) {
969 				fputs(else_defined, ofp);
970 				def_state = IN_ELSE;
971 			}
972 		}
973 		while (new <= pat_end && pch_char(new) == '+') {
974 			fputs(pfetch(new), ofp);
975 			new++;
976 		}
977 	}
978 	if (do_defines && def_state != OUTSIDE) {
979 		fputs(end_defined, ofp);
980 	}
981 }
982 
983 /*
984  * Open the new file.
985  */
986 static void
init_output(const char * name)987 init_output(const char *name)
988 {
989 	ofp = fopen(name, "w");
990 	if (ofp == NULL)
991 		pfatal("can't create %s", name);
992 }
993 
994 /*
995  * Open a file to put hunks we can't locate.
996  */
997 static void
init_reject(const char * name)998 init_reject(const char *name)
999 {
1000 	rejfp = fopen(name, "w");
1001 	if (rejfp == NULL)
1002 		pfatal("can't create %s", name);
1003 }
1004 
1005 /*
1006  * Copy input file to output, up to wherever hunk is to be applied.
1007  * If endoffile is true, treat the last line specially since it may
1008  * lack a newline.
1009  */
1010 static void
copy_till(LINENUM lastline,bool endoffile)1011 copy_till(LINENUM lastline, bool endoffile)
1012 {
1013 	if (last_frozen_line > lastline)
1014 		fatal("misordered hunks! output would be garbled\n");
1015 	while (last_frozen_line < lastline) {
1016 		if (++last_frozen_line == lastline && endoffile)
1017 			dump_line(last_frozen_line, !last_line_missing_eol);
1018 		else
1019 			dump_line(last_frozen_line, true);
1020 	}
1021 }
1022 
1023 /*
1024  * Finish copying the input file to the output file.
1025  */
1026 static void
spew_output(void)1027 spew_output(void)
1028 {
1029 #ifdef DEBUGGING
1030 	if (debug & 256)
1031 		say("il=%ld lfl=%ld\n", input_lines, last_frozen_line);
1032 #endif
1033 	if (input_lines)
1034 		copy_till(input_lines, true);	/* dump remainder of file */
1035 	fclose(ofp);
1036 	ofp = NULL;
1037 }
1038 
1039 /*
1040  * Copy one line from input to output.
1041  */
1042 static void
dump_line(LINENUM line,bool write_newline)1043 dump_line(LINENUM line, bool write_newline)
1044 {
1045 	char	*s;
1046 
1047 	s = ifetch(line, 0);
1048 	if (s == NULL)
1049 		return;
1050 	/* Note: string is not NUL terminated. */
1051 	for (; *s != '\n'; s++)
1052 		putc(*s, ofp);
1053 	if (write_newline)
1054 		putc('\n', ofp);
1055 }
1056 
1057 /*
1058  * Does the patch pattern match at line base+offset?
1059  */
1060 static bool
patch_match(LINENUM base,LINENUM offset,LINENUM fuzz)1061 patch_match(LINENUM base, LINENUM offset, LINENUM fuzz)
1062 {
1063 	LINENUM		pline = 1 + fuzz;
1064 	LINENUM		iline;
1065 	LINENUM		pat_lines = pch_ptrn_lines() - fuzz;
1066 	const char	*ilineptr;
1067 	const char	*plineptr;
1068 	ssize_t		plinelen;
1069 
1070 	for (iline = base + offset + fuzz; pline <= pat_lines; pline++, iline++) {
1071 		ilineptr = ifetch(iline, offset >= 0);
1072 		if (ilineptr == NULL)
1073 			return false;
1074 		plineptr = pfetch(pline);
1075 		plinelen = pch_line_len(pline);
1076 		if (canonicalize) {
1077 			if (!similar(ilineptr, plineptr, plinelen))
1078 				return false;
1079 		} else if (strnNE(ilineptr, plineptr, plinelen))
1080 			return false;
1081 		if (iline == input_lines) {
1082 			/*
1083 			 * We are looking at the last line of the file.
1084 			 * If the file has no eol, the patch line should
1085 			 * not have one either and vice-versa. Note that
1086 			 * plinelen > 0.
1087 			 */
1088 			if (last_line_missing_eol) {
1089 				if (plineptr[plinelen - 1] == '\n')
1090 					return false;
1091 			} else {
1092 				if (plineptr[plinelen - 1] != '\n')
1093 					return false;
1094 			}
1095 		}
1096 	}
1097 	return true;
1098 }
1099 
1100 /*
1101  * Do two lines match with canonicalized white space?
1102  */
1103 static bool
similar(const char * a,const char * b,ssize_t len)1104 similar(const char *a, const char *b, ssize_t len)
1105 {
1106 	while (len) {
1107 		if (isspace((unsigned char)*b)) { /* whitespace (or \n) to match? */
1108 			if (!isspace((unsigned char)*a))
1109 				return false;	/* no corresponding whitespace */
1110 			while (len && isspace((unsigned char)*b) && *b != '\n')
1111 				b++, len--;	/* skip pattern whitespace */
1112 			while (isspace((unsigned char)*a) && *a != '\n')
1113 				a++;	/* skip target whitespace */
1114 			if (*a == '\n' || *b == '\n')
1115 				return (*a == *b);	/* should end in sync */
1116 		} else if (*a++ != *b++)	/* match non-whitespace chars */
1117 			return false;
1118 		else
1119 			len--;	/* probably not necessary */
1120 	}
1121 	return true;		/* actually, this is not reached */
1122 	/* since there is always a \n */
1123 }
1124