xref: /freebsd/usr.bin/diff/diffreg.c (revision 9317242469f1ca682626d9806f8caf65d143c09a)
1 /*	$OpenBSD: diffreg.c,v 1.93 2019/06/28 13:35:00 deraadt Exp $	*/
2 
3 /*-
4  * SPDX-License-Identifier: BSD-4-Clause
5  *
6  * Copyright (C) Caldera International Inc.  2001-2002.
7  * All rights reserved.
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions
11  * are met:
12  * 1. Redistributions of source code and documentation must retain the above
13  *    copyright notice, this list of conditions and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  * 3. All advertising materials mentioning features or use of this software
18  *    must display the following acknowledgement:
19  *	This product includes software developed or owned by Caldera
20  *	International, Inc.
21  * 4. Neither the name of Caldera International, Inc. nor the names of other
22  *    contributors may be used to endorse or promote products derived from
23  *    this software without specific prior written permission.
24  *
25  * USE OF THE SOFTWARE PROVIDED FOR UNDER THIS LICENSE BY CALDERA
26  * INTERNATIONAL, INC. AND CONTRIBUTORS ``AS IS'' AND ANY EXPRESS OR
27  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
28  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
29  * IN NO EVENT SHALL CALDERA INTERNATIONAL, INC. BE LIABLE FOR ANY DIRECT,
30  * INDIRECT INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
31  * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
32  * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
33  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
34  * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
35  * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
36  * POSSIBILITY OF SUCH DAMAGE.
37  */
38 /*-
39  * Copyright (c) 1991, 1993
40  *	The Regents of the University of California.  All rights reserved.
41  *
42  * Redistribution and use in source and binary forms, with or without
43  * modification, are permitted provided that the following conditions
44  * are met:
45  * 1. Redistributions of source code must retain the above copyright
46  *    notice, this list of conditions and the following disclaimer.
47  * 2. Redistributions in binary form must reproduce the above copyright
48  *    notice, this list of conditions and the following disclaimer in the
49  *    documentation and/or other materials provided with the distribution.
50  * 3. Neither the name of the University nor the names of its contributors
51  *    may be used to endorse or promote products derived from this software
52  *    without specific prior written permission.
53  *
54  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
55  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
56  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
57  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
58  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
59  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
60  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
61  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
62  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
63  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
64  * SUCH DAMAGE.
65  */
66 
67 #include <sys/capsicum.h>
68 #include <sys/stat.h>
69 
70 #include <capsicum_helpers.h>
71 #include <ctype.h>
72 #include <err.h>
73 #include <errno.h>
74 #include <fcntl.h>
75 #include <limits.h>
76 #include <math.h>
77 #include <paths.h>
78 #include <regex.h>
79 #include <stdbool.h>
80 #include <stddef.h>
81 #include <stdint.h>
82 #include <stdio.h>
83 #include <stdlib.h>
84 #include <string.h>
85 
86 #include "pr.h"
87 #include "diff.h"
88 #include "xmalloc.h"
89 
90 /*
91  * diff - compare two files.
92  */
93 
94 /*
95  *	Uses an algorithm due to Harold Stone, which finds a pair of longest
96  *	identical subsequences in the two files.
97  *
98  *	The major goal is to generate the match vector J. J[i] is the index of
99  *	the line in file1 corresponding to line i file0. J[i] = 0 if there is no
100  *	such line in file1.
101  *
102  *	Lines are hashed so as to work in core. All potential matches are
103  *	located by sorting the lines of each file on the hash (called
104  *	``value''). In particular, this collects the equivalence classes in
105  *	file1 together. Subroutine equiv replaces the value of each line in
106  *	file0 by the index of the first element of its matching equivalence in
107  *	(the reordered) file1. To save space equiv squeezes file1 into a single
108  *	array member in which the equivalence classes are simply concatenated,
109  *	except that their first members are flagged by changing sign.
110  *
111  *	Next the indices that point into member are unsorted into array class
112  *	according to the original order of file0.
113  *
114  *	The cleverness lies in routine stone. This marches through the lines of
115  *	file0, developing a vector klist of "k-candidates". At step i
116  *	a k-candidate is a matched pair of lines x,y (x in file0 y in file1)
117  *	such that there is a common subsequence of length k between the first
118  *	i lines of file0 and the first y lines of file1, but there is no such
119  *	subsequence for any smaller y. x is the earliest possible mate to y that
120  *	occurs in such a subsequence.
121  *
122  *	Whenever any of the members of the equivalence class of lines in file1
123  *	matable to a line in file0 has serial number less than the y of some
124  *	k-candidate, that k-candidate with the smallest such y is replaced. The
125  *	new k-candidate is chained (via pred) to the current k-1 candidate so
126  *	that the actual subsequence can be recovered. When a member has serial
127  *	number greater that the y of all k-candidates, the klist is extended. At
128  *	the end, the longest subsequence is pulled out and placed in the array J
129  *	by unravel.
130  *
131  *	With J in hand, the matches there recorded are check'ed against reality
132  *	to assure that no spurious matches have crept in due to hashing. If they
133  *	have, they are broken, and "jackpot" is recorded -- a harmless matter
134  *	except that a true match for a spuriously mated line may now be
135  *	unnecessarily reported as a change.
136  *
137  *	Much of the complexity of the program comes simply from trying to
138  *	minimize core utilization and maximize the range of doable problems by
139  *	dynamically allocating what is needed and reusing what is not. The core
140  *	requirements for problems larger than somewhat are (in words)
141  *	2*length(file0) + length(file1) + 3*(number of k-candidates installed),
142  *	typically about 6n words for files of length n.
143  */
144 
145 struct cand {
146 	int	x;
147 	int	y;
148 	int	pred;
149 };
150 
151 static struct line {
152 	int	serial;
153 	int	value;
154 } *file[2];
155 
156 /*
157  * The following struct is used to record change information when
158  * doing a "context" or "unified" diff.  (see routine "change" to
159  * understand the highly mnemonic field names)
160  */
161 struct context_vec {
162 	int	a;		/* start line in old file */
163 	int	b;		/* end line in old file */
164 	int	c;		/* start line in new file */
165 	int	d;		/* end line in new file */
166 };
167 
168 enum readhash { RH_BINARY, RH_OK, RH_EOF };
169 
170 static int	 diffreg_stone(char *, char *, int, int);
171 static FILE	*opentemp(const char *);
172 static void	 output(char *, FILE *, char *, FILE *, int);
173 static void	 check(FILE *, FILE *, int);
174 static void	 range(int, int, const char *);
175 static void	 uni_range(int, int);
176 static void	 dump_context_vec(FILE *, FILE *, int);
177 static void	 dump_unified_vec(FILE *, FILE *, int);
178 static bool	 prepare(int, FILE *, size_t, int);
179 static void	 prune(void);
180 static void	 equiv(struct line *, int, struct line *, int, int *);
181 static void	 unravel(int);
182 static void	 unsort(struct line *, int, int *);
183 static void	 change(char *, FILE *, char *, FILE *, int, int, int, int, int *);
184 static void	 sort(struct line *, int);
185 static void	 print_header(const char *, const char *);
186 static void	 print_space(int, int, int);
187 static bool	 ignoreline_pattern(char *);
188 static bool	 ignoreline(char *, bool);
189 static int	 asciifile(FILE *);
190 static int	 fetch(long *, int, int, FILE *, int, int, int);
191 static int	 newcand(int, int, int);
192 static int	 search(int *, int, int);
193 static int	 skipline(FILE *);
194 static int	 stone(int *, int, int *, int *, int);
195 static enum readhash readhash(FILE *, int, unsigned *);
196 static int	 files_differ(FILE *, FILE *, int);
197 static char	*match_function(const long *, int, FILE *);
198 static char	*preadline(int, size_t, off_t);
199 
200 static int	 *J;			/* will be overlaid on class */
201 static int	 *class;		/* will be overlaid on file[0] */
202 static int	 *klist;		/* will be overlaid on file[0] after class */
203 static int	 *member;		/* will be overlaid on file[1] */
204 static int	 clen;
205 static int	 inifdef;		/* whether or not we are in a #ifdef block */
206 static size_t	 len[2];		/* lengths of files in lines */
207 static size_t	 pref, suff;		/* lengths of prefix and suffix */
208 static size_t	 slen[2];		/* lengths of files minus pref / suff */
209 static int	 anychange;
210 static int	 hw, lpad,rpad;		/* half width and padding */
211 static int	 edoffset;
212 static long	*ixnew;		/* will be overlaid on file[1] */
213 static long	*ixold;		/* will be overlaid on klist */
214 static struct cand *clist;	/* merely a free storage pot for candidates */
215 static int	 clistlen;		/* the length of clist */
216 static struct line *sfile[2];	/* shortened by pruning common prefix/suffix */
217 static int	(*chrtran)(int);	/* translation table for case-folding */
218 static struct context_vec *context_vec_start;
219 static struct context_vec *context_vec_end;
220 static struct context_vec *context_vec_ptr;
221 
222 #define FUNCTION_CONTEXT_SIZE	55
223 static char lastbuf[FUNCTION_CONTEXT_SIZE];
224 static int lastline;
225 static int lastmatchline;
226 
227 int
228 diffreg(char *file1, char *file2, int flags, int capsicum)
229 {
230 	/*
231 	 * If we have set the algorithm with -A or --algorithm use that if we
232 	 * can and if not print an error.
233 	 */
234 	if (diff_algorithm_set) {
235 		if (diff_algorithm == D_DIFFMYERS ||
236 		    diff_algorithm == D_DIFFPATIENCE) {
237 			if (can_libdiff(flags))
238 				return diffreg_new(file1, file2, flags, capsicum);
239 			else
240 				errx(2, "cannot use Myers algorithm with selected options");
241 		} else {
242 			/* Fallback to using stone. */
243 			return diffreg_stone(file1, file2, flags, capsicum);
244 		}
245 	} else {
246 		if (can_libdiff(flags))
247 			return diffreg_new(file1, file2, flags, capsicum);
248 		else
249 			return diffreg_stone(file1, file2, flags, capsicum);
250 	}
251 }
252 
253 static int
254 clow2low(int c)
255 {
256 
257 	return (c);
258 }
259 
260 static int
261 cup2low(int c)
262 {
263 
264 	return (tolower(c));
265 }
266 
267 int
268 diffreg_stone(char *file1, char *file2, int flags, int capsicum)
269 {
270 	FILE *f1, *f2;
271 	int i, rval;
272 	struct pr *pr = NULL;
273 	cap_rights_t rights_ro;
274 
275 	f1 = f2 = NULL;
276 	rval = D_SAME;
277 	anychange = 0;
278 	lastline = 0;
279 	lastmatchline = 0;
280 
281 	/*
282 	 * In side-by-side mode, we need to print the left column, a
283 	 * change marker surrounded by padding, and the right column.
284 	 *
285 	 * If expanding tabs, we don't care about alignment, so we simply
286 	 * subtract 3 from the width and divide by two.
287 	 *
288 	 * If not expanding tabs, we need to ensure that the right column
289 	 * is aligned to a tab stop.  We start with the same formula, then
290 	 * decrement until we reach a size that lets us tab-align the
291 	 * right column.  We then adjust the width down if necessary for
292 	 * the padding calculation to work.
293 	 *
294 	 * Left padding is half the space left over, rounded down; right
295 	 * padding is whatever is needed to match the width.
296 	 */
297 	if (diff_format == D_SIDEBYSIDE) {
298 		if (flags & D_EXPANDTABS) {
299 			if (width > 3) {
300 				hw = (width - 3) / 2;
301 			} else {
302 				/* not enough space */
303 				hw = 0;
304 			}
305 		} else if (width <= 3 || width <= tabsize) {
306 			/* not enough space */
307 			hw = 0;
308 		} else {
309 			hw = (width - 3) / 2;
310 			while (hw > 0 && roundup(hw + 3, tabsize) + hw > width)
311 				hw--;
312 			if (width - (roundup(hw + 3, tabsize) + hw) < tabsize)
313 				width = roundup(hw + 3, tabsize) + hw;
314 		}
315 		lpad = (width - hw * 2 - 1) / 2;
316 		rpad = (width - hw * 2 - 1) - lpad;
317 	}
318 
319 	if (flags & D_IGNORECASE)
320 		chrtran = cup2low;
321 	else
322 		chrtran = clow2low;
323 	if (S_ISDIR(stb1.st_mode) != S_ISDIR(stb2.st_mode))
324 		return (S_ISDIR(stb1.st_mode) ? D_MISMATCH1 : D_MISMATCH2);
325 	if (strcmp(file1, "-") == 0 && strcmp(file2, "-") == 0)
326 		goto closem;
327 
328 	if (flags & D_EMPTY1)
329 		f1 = fopen(_PATH_DEVNULL, "r");
330 	else {
331 		if (!S_ISREG(stb1.st_mode)) {
332 			if ((f1 = opentemp(file1)) == NULL ||
333 			    fstat(fileno(f1), &stb1) == -1) {
334 				warn("%s", file1);
335 				rval = D_ERROR;
336 				status |= 2;
337 				goto closem;
338 			}
339 		} else if (strcmp(file1, "-") == 0)
340 			f1 = stdin;
341 		else
342 			f1 = fopen(file1, "r");
343 	}
344 	if (f1 == NULL) {
345 		warn("%s", file1);
346 		rval = D_ERROR;
347 		status |= 2;
348 		goto closem;
349 	}
350 
351 	if (flags & D_EMPTY2)
352 		f2 = fopen(_PATH_DEVNULL, "r");
353 	else {
354 		if (!S_ISREG(stb2.st_mode)) {
355 			if ((f2 = opentemp(file2)) == NULL ||
356 			    fstat(fileno(f2), &stb2) == -1) {
357 				warn("%s", file2);
358 				rval = D_ERROR;
359 				status |= 2;
360 				goto closem;
361 			}
362 		} else if (strcmp(file2, "-") == 0)
363 			f2 = stdin;
364 		else
365 			f2 = fopen(file2, "r");
366 	}
367 	if (f2 == NULL) {
368 		warn("%s", file2);
369 		rval = D_ERROR;
370 		status |= 2;
371 		goto closem;
372 	}
373 
374 	if (lflag)
375 		pr = start_pr(file1, file2);
376 
377 	if (capsicum) {
378 		cap_rights_init(&rights_ro, CAP_READ, CAP_FSTAT, CAP_SEEK);
379 		if (caph_rights_limit(fileno(f1), &rights_ro) < 0)
380 			err(2, "unable to limit rights on: %s", file1);
381 		if (caph_rights_limit(fileno(f2), &rights_ro) < 0)
382 			err(2, "unable to limit rights on: %s", file2);
383 		if (fileno(f1) == STDIN_FILENO || fileno(f2) == STDIN_FILENO) {
384 			/* stdin has already been limited */
385 			if (caph_limit_stderr() == -1)
386 				err(2, "unable to limit stderr");
387 			if (caph_limit_stdout() == -1)
388 				err(2, "unable to limit stdout");
389 		} else if (caph_limit_stdio() == -1)
390 				err(2, "unable to limit stdio");
391 
392 		caph_cache_catpages();
393 		caph_cache_tzdata();
394 		if (caph_enter() < 0)
395 			err(2, "unable to enter capability mode");
396 	}
397 
398 	switch (files_differ(f1, f2, flags)) {
399 	case 0:
400 		goto closem;
401 	case 1:
402 		break;
403 	default:
404 		/* error */
405 		rval = D_ERROR;
406 		status |= 2;
407 		goto closem;
408 	}
409 
410 	if (diff_format == D_BRIEF && ignore_pats == NULL &&
411 	    (flags & (D_FOLDBLANKS|D_IGNOREBLANKS|D_IGNORECASE|
412 	    D_SKIPBLANKLINES|D_STRIPCR)) == 0)
413 	{
414 		rval = D_DIFFER;
415 		status |= 1;
416 		goto closem;
417 	}
418 	if ((flags & D_FORCEASCII) != 0) {
419 		(void)prepare(0, f1, stb1.st_size, flags);
420 		(void)prepare(1, f2, stb2.st_size, flags);
421 	} else if (!asciifile(f1) || !asciifile(f2) ||
422 		    !prepare(0, f1, stb1.st_size, flags) ||
423 		    !prepare(1, f2, stb2.st_size, flags)) {
424 		rval = D_BINARY;
425 		status |= 1;
426 		goto closem;
427 	}
428 	if (len[0] > INT_MAX - 2)
429 		errc(1, EFBIG, "%s", file1);
430 	if (len[1] > INT_MAX - 2)
431 		errc(1, EFBIG, "%s", file2);
432 
433 	prune();
434 	sort(sfile[0], slen[0]);
435 	sort(sfile[1], slen[1]);
436 
437 	member = (int *)file[1];
438 	equiv(sfile[0], slen[0], sfile[1], slen[1], member);
439 	member = xreallocarray(member, slen[1] + 2, sizeof(*member));
440 
441 	class = (int *)file[0];
442 	unsort(sfile[0], slen[0], class);
443 	class = xreallocarray(class, slen[0] + 2, sizeof(*class));
444 
445 	klist = xcalloc(slen[0] + 2, sizeof(*klist));
446 	clen = 0;
447 	clistlen = 100;
448 	clist = xcalloc(clistlen, sizeof(*clist));
449 	i = stone(class, slen[0], member, klist, flags);
450 	free(member);
451 	free(class);
452 
453 	J = xreallocarray(J, len[0] + 2, sizeof(*J));
454 	unravel(klist[i]);
455 	free(clist);
456 	free(klist);
457 
458 	ixold = xreallocarray(ixold, len[0] + 2, sizeof(*ixold));
459 	ixnew = xreallocarray(ixnew, len[1] + 2, sizeof(*ixnew));
460 	check(f1, f2, flags);
461 	output(file1, f1, file2, f2, flags);
462 
463 closem:
464 	if (pr != NULL)
465 		stop_pr(pr);
466 	if (anychange) {
467 		status |= 1;
468 		if (rval == D_SAME)
469 			rval = D_DIFFER;
470 	}
471 	if (f1 != NULL)
472 		fclose(f1);
473 	if (f2 != NULL)
474 		fclose(f2);
475 
476 	return (rval);
477 }
478 
479 /*
480  * Check to see if the given files differ.
481  * Returns 0 if they are the same, 1 if different, and -1 on error.
482  * XXX - could use code from cmp(1) [faster]
483  */
484 static int
485 files_differ(FILE *f1, FILE *f2, int flags)
486 {
487 	char buf1[BUFSIZ], buf2[BUFSIZ];
488 	size_t i, j;
489 
490 	if ((flags & (D_EMPTY1|D_EMPTY2)) || stb1.st_size != stb2.st_size ||
491 	    (stb1.st_mode & S_IFMT) != (stb2.st_mode & S_IFMT))
492 		return (1);
493 
494 	if (stb1.st_dev == stb2.st_dev && stb1.st_ino == stb2.st_ino)
495 		return (0);
496 
497 	for (;;) {
498 		i = fread(buf1, 1, sizeof(buf1), f1);
499 		j = fread(buf2, 1, sizeof(buf2), f2);
500 		if ((!i && ferror(f1)) || (!j && ferror(f2)))
501 			return (-1);
502 		if (i != j)
503 			return (1);
504 		if (i == 0)
505 			return (0);
506 		if (memcmp(buf1, buf2, i) != 0)
507 			return (1);
508 	}
509 }
510 
511 static FILE *
512 opentemp(const char *f)
513 {
514 	char buf[BUFSIZ], tempfile[PATH_MAX];
515 	ssize_t nread;
516 	int ifd, ofd;
517 
518 	if (strcmp(f, "-") == 0)
519 		ifd = STDIN_FILENO;
520 	else if ((ifd = open(f, O_RDONLY, 0644)) == -1)
521 		return (NULL);
522 
523 	(void)strlcpy(tempfile, _PATH_TMP "/diff.XXXXXXXX", sizeof(tempfile));
524 
525 	if ((ofd = mkstemp(tempfile)) == -1) {
526 		close(ifd);
527 		return (NULL);
528 	}
529 	unlink(tempfile);
530 	while ((nread = read(ifd, buf, BUFSIZ)) > 0) {
531 		if (write(ofd, buf, nread) != nread) {
532 			close(ifd);
533 			close(ofd);
534 			return (NULL);
535 		}
536 	}
537 	close(ifd);
538 	lseek(ofd, (off_t)0, SEEK_SET);
539 	return (fdopen(ofd, "r"));
540 }
541 
542 static bool
543 prepare(int i, FILE *fd, size_t filesize, int flags)
544 {
545 	struct line *p;
546 	unsigned h;
547 	size_t sz, j = 0;
548 	enum readhash r;
549 
550 	rewind(fd);
551 
552 	sz = MIN(filesize, SIZE_MAX) / 25;
553 	if (sz < 100)
554 		sz = 100;
555 
556 	p = xcalloc(sz + 3, sizeof(*p));
557 	while ((r = readhash(fd, flags, &h)) != RH_EOF) {
558 		if (r == RH_BINARY)
559 			return (false);
560 		if (j == SIZE_MAX)
561 			break;
562 		if (j == sz) {
563 			sz = sz * 3 / 2;
564 			p = xreallocarray(p, sz + 3, sizeof(*p));
565 		}
566 		p[++j].value = h;
567 	}
568 
569 	len[i] = j;
570 	file[i] = p;
571 
572 	return (true);
573 }
574 
575 static void
576 prune(void)
577 {
578 	size_t i, j;
579 
580 	for (pref = 0; pref < len[0] && pref < len[1] &&
581 	    file[0][pref + 1].value == file[1][pref + 1].value;
582 	    pref++)
583 		;
584 	for (suff = 0; suff < len[0] - pref && suff < len[1] - pref &&
585 	    file[0][len[0] - suff].value == file[1][len[1] - suff].value;
586 	    suff++)
587 		;
588 	for (j = 0; j < 2; j++) {
589 		sfile[j] = file[j] + pref;
590 		slen[j] = len[j] - pref - suff;
591 		for (i = 0; i <= slen[j]; i++)
592 			sfile[j][i].serial = i;
593 	}
594 }
595 
596 static void
597 equiv(struct line *a, int n, struct line *b, int m, int *c)
598 {
599 	int i, j;
600 
601 	i = j = 1;
602 	while (i <= n && j <= m) {
603 		if (a[i].value < b[j].value)
604 			a[i++].value = 0;
605 		else if (a[i].value == b[j].value)
606 			a[i++].value = j;
607 		else
608 			j++;
609 	}
610 	while (i <= n)
611 		a[i++].value = 0;
612 	b[m + 1].value = 0;
613 	j = 0;
614 	while (++j <= m) {
615 		c[j] = -b[j].serial;
616 		while (b[j + 1].value == b[j].value) {
617 			j++;
618 			c[j] = b[j].serial;
619 		}
620 	}
621 	c[j] = -1;
622 }
623 
624 static int
625 stone(int *a, int n, int *b, int *c, int flags)
626 {
627 	int i, k, y, j, l;
628 	int oldc, tc, oldl, sq;
629 	unsigned numtries, bound;
630 
631 	if (flags & D_MINIMAL)
632 		bound = UINT_MAX;
633 	else {
634 		sq = sqrt(n);
635 		bound = MAX(256, sq);
636 	}
637 
638 	k = 0;
639 	c[0] = newcand(0, 0, 0);
640 	for (i = 1; i <= n; i++) {
641 		j = a[i];
642 		if (j == 0)
643 			continue;
644 		y = -b[j];
645 		oldl = 0;
646 		oldc = c[0];
647 		numtries = 0;
648 		do {
649 			if (y <= clist[oldc].y)
650 				continue;
651 			l = search(c, k, y);
652 			if (l != oldl + 1)
653 				oldc = c[l - 1];
654 			if (l <= k) {
655 				if (clist[c[l]].y <= y)
656 					continue;
657 				tc = c[l];
658 				c[l] = newcand(i, y, oldc);
659 				oldc = tc;
660 				oldl = l;
661 				numtries++;
662 			} else {
663 				c[l] = newcand(i, y, oldc);
664 				k++;
665 				break;
666 			}
667 		} while ((y = b[++j]) > 0 && numtries < bound);
668 	}
669 	return (k);
670 }
671 
672 static int
673 newcand(int x, int y, int pred)
674 {
675 	struct cand *q;
676 
677 	if (clen == clistlen) {
678 		clistlen = clistlen * 11 / 10;
679 		clist = xreallocarray(clist, clistlen, sizeof(*clist));
680 	}
681 	q = clist + clen;
682 	q->x = x;
683 	q->y = y;
684 	q->pred = pred;
685 	return (clen++);
686 }
687 
688 static int
689 search(int *c, int k, int y)
690 {
691 	int i, j, l, t;
692 
693 	if (clist[c[k]].y < y)	/* quick look for typical case */
694 		return (k + 1);
695 	i = 0;
696 	j = k + 1;
697 	for (;;) {
698 		l = (i + j) / 2;
699 		if (l <= i)
700 			break;
701 		t = clist[c[l]].y;
702 		if (t > y)
703 			j = l;
704 		else if (t < y)
705 			i = l;
706 		else
707 			return (l);
708 	}
709 	return (l + 1);
710 }
711 
712 static void
713 unravel(int p)
714 {
715 	struct cand *q;
716 	size_t i;
717 
718 	for (i = 0; i <= len[0]; i++)
719 		J[i] = i <= pref ? i :
720 		    i > len[0] - suff ? i + len[1] - len[0] : 0;
721 	for (q = clist + p; q->y != 0; q = clist + q->pred)
722 		J[q->x + pref] = q->y + pref;
723 }
724 
725 /*
726  * Check does double duty:
727  *  1. ferret out any fortuitous correspondences due to confounding by
728  *     hashing (which result in "jackpot")
729  *  2. collect random access indexes to the two files
730  */
731 static void
732 check(FILE *f1, FILE *f2, int flags)
733 {
734 	int i, j, /* jackpot, */ c, d;
735 	long ctold, ctnew;
736 
737 	rewind(f1);
738 	rewind(f2);
739 	j = 1;
740 	ixold[0] = ixnew[0] = 0;
741 	/* jackpot = 0; */
742 	ctold = ctnew = 0;
743 	for (i = 1; i <= (int)len[0]; i++) {
744 		if (J[i] == 0) {
745 			ixold[i] = ctold += skipline(f1);
746 			continue;
747 		}
748 		while (j < J[i]) {
749 			ixnew[j] = ctnew += skipline(f2);
750 			j++;
751 		}
752 		if (flags & (D_FOLDBLANKS | D_IGNOREBLANKS | D_IGNORECASE | D_STRIPCR)) {
753 			for (;;) {
754 				c = getc(f1);
755 				d = getc(f2);
756 				/*
757 				 * GNU diff ignores a missing newline
758 				 * in one file for -b or -w.
759 				 */
760 				if (flags & (D_FOLDBLANKS | D_IGNOREBLANKS)) {
761 					if (c == EOF && isspace(d)) {
762 						ctnew++;
763 						break;
764 					} else if (isspace(c) && d == EOF) {
765 						ctold++;
766 						break;
767 					}
768 				}
769 				ctold++;
770 				ctnew++;
771 				if (flags & D_STRIPCR && (c == '\r' || d == '\r')) {
772 					if (c == '\r') {
773 						if ((c = getc(f1)) == '\n') {
774 							ctold++;
775 						} else {
776 							ungetc(c, f1);
777 						}
778 					}
779 					if (d == '\r') {
780 						if ((d = getc(f2)) == '\n') {
781 							ctnew++;
782 						} else {
783 							ungetc(d, f2);
784 						}
785 					}
786 					break;
787 				}
788 				if ((flags & D_FOLDBLANKS) && isspace(c) &&
789 				    isspace(d)) {
790 					do {
791 						if (c == '\n')
792 							break;
793 						ctold++;
794 					} while (isspace(c = getc(f1)));
795 					do {
796 						if (d == '\n')
797 							break;
798 						ctnew++;
799 					} while (isspace(d = getc(f2)));
800 				} else if (flags & D_IGNOREBLANKS) {
801 					while (isspace(c) && c != '\n') {
802 						c = getc(f1);
803 						ctold++;
804 					}
805 					while (isspace(d) && d != '\n') {
806 						d = getc(f2);
807 						ctnew++;
808 					}
809 				}
810 				if (chrtran(c) != chrtran(d)) {
811 					/* jackpot++; */
812 					J[i] = 0;
813 					if (c != '\n' && c != EOF)
814 						ctold += skipline(f1);
815 					if (d != '\n' && c != EOF)
816 						ctnew += skipline(f2);
817 					break;
818 				}
819 				if (c == '\n' || c == EOF)
820 					break;
821 			}
822 		} else {
823 			for (;;) {
824 				ctold++;
825 				ctnew++;
826 				if ((c = getc(f1)) != (d = getc(f2))) {
827 					/* jackpot++; */
828 					J[i] = 0;
829 					if (c != '\n' && c != EOF)
830 						ctold += skipline(f1);
831 					if (d != '\n' && c != EOF)
832 						ctnew += skipline(f2);
833 					break;
834 				}
835 				if (c == '\n' || c == EOF)
836 					break;
837 			}
838 		}
839 		ixold[i] = ctold;
840 		ixnew[j] = ctnew;
841 		j++;
842 	}
843 	for (; j <= (int)len[1]; j++) {
844 		ixnew[j] = ctnew += skipline(f2);
845 	}
846 	/*
847 	 * if (jackpot)
848 	 *	fprintf(stderr, "jackpot\n");
849 	 */
850 }
851 
852 /* shellsort CACM #201 */
853 static void
854 sort(struct line *a, int n)
855 {
856 	struct line *ai, *aim, w;
857 	int j, m = 0, k;
858 
859 	if (n == 0)
860 		return;
861 	for (j = 1; j <= n; j *= 2)
862 		m = 2 * j - 1;
863 	for (m /= 2; m != 0; m /= 2) {
864 		k = n - m;
865 		for (j = 1; j <= k; j++) {
866 			for (ai = &a[j]; ai > a; ai -= m) {
867 				aim = &ai[m];
868 				if (aim < ai)
869 					break;	/* wraparound */
870 				if (aim->value > ai[0].value ||
871 				    (aim->value == ai[0].value &&
872 					aim->serial > ai[0].serial))
873 					break;
874 				w.value = ai[0].value;
875 				ai[0].value = aim->value;
876 				aim->value = w.value;
877 				w.serial = ai[0].serial;
878 				ai[0].serial = aim->serial;
879 				aim->serial = w.serial;
880 			}
881 		}
882 	}
883 }
884 
885 static void
886 unsort(struct line *f, int l, int *b)
887 {
888 	int *a, i;
889 
890 	a = xcalloc(l + 1, sizeof(*a));
891 	for (i = 1; i <= l; i++)
892 		a[f[i].serial] = f[i].value;
893 	for (i = 1; i <= l; i++)
894 		b[i] = a[i];
895 	free(a);
896 }
897 
898 static int
899 skipline(FILE *f)
900 {
901 	int i, c;
902 
903 	for (i = 1; (c = getc(f)) != '\n' && c != EOF; i++)
904 		continue;
905 	return (i);
906 }
907 
908 static void
909 output(char *file1, FILE *f1, char *file2, FILE *f2, int flags)
910 {
911 	int i, j, m, i0, i1, j0, j1, nc;
912 
913 	rewind(f1);
914 	rewind(f2);
915 	m = len[0];
916 	J[0] = 0;
917 	J[m + 1] = len[1] + 1;
918 	if (diff_format != D_EDIT) {
919 		for (i0 = 1; i0 <= m; i0 = i1 + 1) {
920 			while (i0 <= m && J[i0] == J[i0 - 1] + 1) {
921 				if (diff_format == D_SIDEBYSIDE && suppress_common != 1) {
922 					nc = fetch(ixold, i0, i0, f1, '\0', 1, flags);
923 					print_space(nc, hw - nc + lpad + 1 + rpad, flags);
924 					fetch(ixnew, J[i0], J[i0], f2, '\0', 0, flags);
925 					printf("\n");
926 				}
927 				i0++;
928 			}
929 			j0 = J[i0 - 1] + 1;
930 			i1 = i0 - 1;
931 			while (i1 < m && J[i1 + 1] == 0)
932 				i1++;
933 			j1 = J[i1 + 1] - 1;
934 			J[i1] = j1;
935 
936 			/*
937 			 * When using side-by-side, lines from both of the files are
938 			 * printed. The algorithm used by diff(1) identifies the ranges
939 			 * in which two files differ.
940 			 * See the change() function below.
941 			 * The for loop below consumes the shorter range, whereas one of
942 			 * the while loops deals with the longer one.
943 			 */
944 			if (diff_format == D_SIDEBYSIDE) {
945 				for (i = i0, j = j0; i <= i1 && j <= j1; i++, j++)
946 					change(file1, f1, file2, f2, i, i, j, j, &flags);
947 
948 				while (i <= i1) {
949 					change(file1, f1, file2, f2, i, i, j + 1, j, &flags);
950 					i++;
951 				}
952 
953 				while (j <= j1) {
954 					change(file1, f1, file2, f2, i + 1, i, j, j, &flags);
955 					j++;
956 				}
957 			} else
958 				change(file1, f1, file2, f2, i0, i1, j0, j1, &flags);
959 		}
960 	} else {
961 		for (i0 = m; i0 >= 1; i0 = i1 - 1) {
962 			while (i0 >= 1 && J[i0] == J[i0 + 1] - 1 && J[i0] != 0)
963 				i0--;
964 			j0 = J[i0 + 1] - 1;
965 			i1 = i0 + 1;
966 			while (i1 > 1 && J[i1 - 1] == 0)
967 				i1--;
968 			j1 = J[i1 - 1] + 1;
969 			J[i1] = j1;
970 			change(file1, f1, file2, f2, i1, i0, j1, j0, &flags);
971 		}
972 	}
973 	if (m == 0)
974 		change(file1, f1, file2, f2, 1, 0, 1, len[1], &flags);
975 	if (diff_format == D_IFDEF || diff_format == D_GFORMAT) {
976 		for (;;) {
977 #define	c i0
978 			if ((c = getc(f1)) == EOF)
979 				return;
980 			printf("%c", c);
981 		}
982 #undef c
983 	}
984 	if (anychange != 0) {
985 		if (diff_format == D_CONTEXT)
986 			dump_context_vec(f1, f2, flags);
987 		else if (diff_format == D_UNIFIED)
988 			dump_unified_vec(f1, f2, flags);
989 	}
990 }
991 
992 static void
993 range(int a, int b, const char *separator)
994 {
995 	printf("%d", a > b ? b : a);
996 	if (a < b)
997 		printf("%s%d", separator, b);
998 }
999 
1000 static void
1001 uni_range(int a, int b)
1002 {
1003 	if (a < b)
1004 		printf("%d,%d", a, b - a + 1);
1005 	else if (a == b)
1006 		printf("%d", b);
1007 	else
1008 		printf("%d,0", b);
1009 }
1010 
1011 static char *
1012 preadline(int fd, size_t rlen, off_t off)
1013 {
1014 	char *line;
1015 	ssize_t nr;
1016 
1017 	line = xmalloc(rlen + 1);
1018 	if ((nr = pread(fd, line, rlen, off)) == -1)
1019 		err(2, "preadline");
1020 	if (nr > 0 && line[nr-1] == '\n')
1021 		nr--;
1022 	line[nr] = '\0';
1023 	return (line);
1024 }
1025 
1026 static bool
1027 ignoreline_pattern(char *line)
1028 {
1029 	int ret;
1030 
1031 	ret = regexec(&ignore_re, line, 0, NULL, 0);
1032 	return (ret == 0);	/* if it matched, it should be ignored. */
1033 }
1034 
1035 static bool
1036 ignoreline(char *line, bool skip_blanks)
1037 {
1038 
1039 	if (skip_blanks && *line == '\0')
1040 		return (true);
1041 	if (ignore_pats != NULL && ignoreline_pattern(line))
1042 		return (true);
1043 	return (false);
1044 }
1045 
1046 /*
1047  * Indicate that there is a difference between lines a and b of the from file
1048  * to get to lines c to d of the to file.  If a is greater then b then there
1049  * are no lines in the from file involved and this means that there were
1050  * lines appended (beginning at b).  If c is greater than d then there are
1051  * lines missing from the to file.
1052  */
1053 static void
1054 change(char *file1, FILE *f1, char *file2, FILE *f2, int a, int b, int c, int d,
1055     int *pflags)
1056 {
1057 	static size_t max_context = 64;
1058 	long curpos;
1059 	int i, nc;
1060 	const char *walk;
1061 	bool skip_blanks, ignore;
1062 
1063 	skip_blanks = (*pflags & D_SKIPBLANKLINES);
1064 restart:
1065 	if ((diff_format != D_IFDEF || diff_format == D_GFORMAT) &&
1066 	    a > b && c > d)
1067 		return;
1068 	if (ignore_pats != NULL || skip_blanks) {
1069 		char *line;
1070 		/*
1071 		 * All lines in the change, insert, or delete must match an ignore
1072 		 * pattern for the change to be ignored.
1073 		 */
1074 		if (a <= b) {		/* Changes and deletes. */
1075 			for (i = a; i <= b; i++) {
1076 				line = preadline(fileno(f1),
1077 				    ixold[i] - ixold[i - 1], ixold[i - 1]);
1078 				ignore = ignoreline(line, skip_blanks);
1079 				free(line);
1080 				if (!ignore)
1081 					goto proceed;
1082 			}
1083 		}
1084 		if (a > b || c <= d) {	/* Changes and inserts. */
1085 			for (i = c; i <= d; i++) {
1086 				line = preadline(fileno(f2),
1087 				    ixnew[i] - ixnew[i - 1], ixnew[i - 1]);
1088 				ignore = ignoreline(line, skip_blanks);
1089 				free(line);
1090 				if (!ignore)
1091 					goto proceed;
1092 			}
1093 		}
1094 		return;
1095 	}
1096 proceed:
1097 	if (*pflags & D_HEADER && diff_format != D_BRIEF) {
1098 		printf("%s %s %s\n", diffargs, file1, file2);
1099 		*pflags &= ~D_HEADER;
1100 	}
1101 	if (diff_format == D_CONTEXT || diff_format == D_UNIFIED) {
1102 		/*
1103 		 * Allocate change records as needed.
1104 		 */
1105 		if (context_vec_start == NULL ||
1106 		    context_vec_ptr == context_vec_end - 1) {
1107 			ptrdiff_t offset = -1;
1108 
1109 			if (context_vec_start != NULL)
1110 				offset = context_vec_ptr - context_vec_start;
1111 			max_context <<= 1;
1112 			context_vec_start = xreallocarray(context_vec_start,
1113 			    max_context, sizeof(*context_vec_start));
1114 			context_vec_end = context_vec_start + max_context;
1115 			context_vec_ptr = context_vec_start + offset;
1116 		}
1117 		if (anychange == 0) {
1118 			/*
1119 			 * Print the context/unidiff header first time through.
1120 			 */
1121 			print_header(file1, file2);
1122 			anychange = 1;
1123 		} else if (a > context_vec_ptr->b + (2 * diff_context) + 1 &&
1124 		    c > context_vec_ptr->d + (2 * diff_context) + 1) {
1125 			/*
1126 			 * If this change is more than 'diff_context' lines from the
1127 			 * previous change, dump the record and reset it.
1128 			 */
1129 			if (diff_format == D_CONTEXT)
1130 				dump_context_vec(f1, f2, *pflags);
1131 			else
1132 				dump_unified_vec(f1, f2, *pflags);
1133 		}
1134 		context_vec_ptr++;
1135 		context_vec_ptr->a = a;
1136 		context_vec_ptr->b = b;
1137 		context_vec_ptr->c = c;
1138 		context_vec_ptr->d = d;
1139 		return;
1140 	}
1141 	if (anychange == 0)
1142 		anychange = 1;
1143 	switch (diff_format) {
1144 	case D_BRIEF:
1145 		return;
1146 	case D_NORMAL:
1147 	case D_EDIT:
1148 		range(a, b, ",");
1149 		printf("%c", a > b ? 'a' : c > d ? 'd' : 'c');
1150 		if (diff_format == D_NORMAL)
1151 			range(c, d, ",");
1152 		printf("\n");
1153 		break;
1154 	case D_REVERSE:
1155 		printf("%c", a > b ? 'a' : c > d ? 'd' : 'c');
1156 		range(a, b, " ");
1157 		printf("\n");
1158 		break;
1159 	case D_NREVERSE:
1160 		if (a > b)
1161 			printf("a%d %d\n", b, d - c + 1);
1162 		else {
1163 			printf("d%d %d\n", a, b - a + 1);
1164 			if (!(c > d))
1165 				/* add changed lines */
1166 				printf("a%d %d\n", b, d - c + 1);
1167 		}
1168 		break;
1169 	}
1170 	if (diff_format == D_GFORMAT) {
1171 		curpos = ftell(f1);
1172 		/* print through if append (a>b), else to (nb: 0 vs 1 orig) */
1173 		nc = ixold[a > b ? b : a - 1] - curpos;
1174 		for (i = 0; i < nc; i++)
1175 			printf("%c", getc(f1));
1176 		for (walk = group_format; *walk != '\0'; walk++) {
1177 			if (*walk == '%') {
1178 				walk++;
1179 				switch (*walk) {
1180 				case '<':
1181 					fetch(ixold, a, b, f1, '<', 1, *pflags);
1182 					break;
1183 				case '>':
1184 					fetch(ixnew, c, d, f2, '>', 0, *pflags);
1185 					break;
1186 				default:
1187 					printf("%%%c", *walk);
1188 					break;
1189 				}
1190 				continue;
1191 			}
1192 			printf("%c", *walk);
1193 		}
1194 	}
1195 	if (diff_format == D_SIDEBYSIDE) {
1196 		if (color && a > b)
1197 			printf("\033[%sm", add_code);
1198 		else if (color && c > d)
1199 			printf("\033[%sm", del_code);
1200 		if (a > b) {
1201 			print_space(0, hw + lpad, *pflags);
1202 		} else {
1203 			nc = fetch(ixold, a, b, f1, '\0', 1, *pflags);
1204 			print_space(nc, hw - nc + lpad, *pflags);
1205 		}
1206 		if (color && a > b)
1207 			printf("\033[%sm", add_code);
1208 		else if (color && c > d)
1209 			printf("\033[%sm", del_code);
1210 		printf("%c", (a > b) ? '>' : ((c > d) ? '<' : '|'));
1211 		if (color && c > d)
1212 			printf("\033[m");
1213 		print_space(hw + lpad + 1, rpad, *pflags);
1214 		fetch(ixnew, c, d, f2, '\0', 0, *pflags);
1215 		printf("\n");
1216 	}
1217 	if (diff_format == D_NORMAL || diff_format == D_IFDEF) {
1218 		fetch(ixold, a, b, f1, '<', 1, *pflags);
1219 		if (a <= b && c <= d && diff_format == D_NORMAL)
1220 			printf("---\n");
1221 	}
1222 	if (diff_format != D_GFORMAT && diff_format != D_SIDEBYSIDE)
1223 		fetch(ixnew, c, d, f2, diff_format == D_NORMAL ? '>' : '\0', 0, *pflags);
1224 	if (edoffset != 0 && diff_format == D_EDIT) {
1225 		/*
1226 		 * A non-zero edoffset value for D_EDIT indicates that the last line
1227 		 * printed was a bare dot (".") that has been escaped as ".." to
1228 		 * prevent ed(1) from misinterpreting it.  We have to add a
1229 		 * substitute command to change this back and restart where we left
1230 		 * off.
1231 		 */
1232 		printf(".\n");
1233 		printf("%ds/.//\n", a + edoffset - 1);
1234 		b = a + edoffset - 1;
1235 		a = b + 1;
1236 		c += edoffset;
1237 		goto restart;
1238 	}
1239 	if ((diff_format == D_EDIT || diff_format == D_REVERSE) && c <= d)
1240 		printf(".\n");
1241 	if (inifdef) {
1242 		printf("#endif /* %s */\n", ifdefname);
1243 		inifdef = 0;
1244 	}
1245 }
1246 
1247 static int
1248 fetch(long *f, int a, int b, FILE *lb, int ch, int oldfile, int flags)
1249 {
1250 	int i, j, c, lastc, col, nc, newcol;
1251 
1252 	edoffset = 0;
1253 	nc = 0;
1254 	col = 0;
1255 	/*
1256 	 * When doing #ifdef's, copy down to current line
1257 	 * if this is the first file, so that stuff makes it to output.
1258 	 */
1259 	if ((diff_format == D_IFDEF) && oldfile) {
1260 		long curpos = ftell(lb);
1261 		/* print through if append (a>b), else to (nb: 0 vs 1 orig) */
1262 		nc = f[a > b ? b : a - 1] - curpos;
1263 		for (i = 0; i < nc; i++)
1264 			printf("%c", getc(lb));
1265 	}
1266 	if (a > b)
1267 		return (0);
1268 	if (diff_format == D_IFDEF) {
1269 		if (inifdef) {
1270 			printf("#else /* %s%s */\n",
1271 			    oldfile == 1 ? "!" : "", ifdefname);
1272 		} else {
1273 			if (oldfile)
1274 				printf("#ifndef %s\n", ifdefname);
1275 			else
1276 				printf("#ifdef %s\n", ifdefname);
1277 		}
1278 		inifdef = 1 + oldfile;
1279 	}
1280 	for (i = a; i <= b; i++) {
1281 		fseek(lb, f[i - 1], SEEK_SET);
1282 		nc = f[i] - f[i - 1];
1283 		if (diff_format == D_SIDEBYSIDE && hw < nc)
1284 			nc = hw;
1285 		if (diff_format != D_IFDEF && diff_format != D_GFORMAT &&
1286 		    ch != '\0') {
1287 			if (color && (ch == '>' || ch == '+'))
1288 				printf("\033[%sm", add_code);
1289 			else if (color && (ch == '<' || ch == '-'))
1290 				printf("\033[%sm", del_code);
1291 			printf("%c", ch);
1292 			if (Tflag && (diff_format == D_NORMAL ||
1293 			    diff_format == D_CONTEXT ||
1294 			    diff_format == D_UNIFIED))
1295 				printf("\t");
1296 			else if (diff_format != D_UNIFIED)
1297 				printf(" ");
1298 		}
1299 		col = j = 0;
1300 		lastc = '\0';
1301 		while (j < nc && (hw == 0 || col < hw)) {
1302 			c = getc(lb);
1303 			if (flags & D_STRIPCR && c == '\r') {
1304 				if ((c = getc(lb)) == '\n')
1305 					j++;
1306 				else {
1307 					ungetc(c, lb);
1308 					c = '\r';
1309 				}
1310 			}
1311 			if (c == EOF) {
1312 				if (diff_format == D_EDIT ||
1313 				    diff_format == D_REVERSE ||
1314 				    diff_format == D_NREVERSE)
1315 					warnx("No newline at end of file");
1316 				else
1317 					printf("\n\\ No newline at end of file\n");
1318 				return (col);
1319 			}
1320 			/*
1321 			 * when using --side-by-side, col needs to be increased
1322 			 * in any case to keep the columns aligned
1323 			 */
1324 			if (c == '\t') {
1325 				/*
1326 				 * Calculate where the tab would bring us.
1327 				 * If it would take us to the end of the
1328 				 * column, either clip it (if expanding
1329 				 * tabs) or return right away (if not).
1330 				 */
1331 				newcol = roundup(col + 1, tabsize);
1332 				if ((flags & D_EXPANDTABS) == 0) {
1333 					if (hw > 0 && newcol >= hw)
1334 						return (col);
1335 					printf("\t");
1336 				} else {
1337 					if (hw > 0 && newcol > hw)
1338 						newcol = hw;
1339 					printf("%*s", newcol - col, "");
1340 				}
1341 				col = newcol;
1342 			} else {
1343 				if (diff_format == D_EDIT && j == 1 && c == '\n' &&
1344 				    lastc == '.') {
1345 					/*
1346 					 * Don't print a bare "." line since that will confuse
1347 					 * ed(1). Print ".." instead and set the, global variable
1348 					 * edoffset to an offset from which to restart. The
1349 					 * caller must check the value of edoffset
1350 					 */
1351 					printf(".\n");
1352 					edoffset = i - a + 1;
1353 					return (edoffset);
1354 				}
1355 				/* when side-by-side, do not print a newline */
1356 				if (diff_format != D_SIDEBYSIDE || c != '\n') {
1357 					if (color && c == '\n')
1358 						printf("\033[m%c", c);
1359 					else
1360 						printf("%c", c);
1361 					col++;
1362 				}
1363 			}
1364 
1365 			j++;
1366 			lastc = c;
1367 		}
1368 	}
1369 	if (color && diff_format == D_SIDEBYSIDE)
1370 		printf("\033[m");
1371 	return (col);
1372 }
1373 
1374 /*
1375  * Hash function taken from Robert Sedgewick, Algorithms in C, 3d ed., p 578.
1376  */
1377 static enum readhash
1378 readhash(FILE *f, int flags, unsigned *hash)
1379 {
1380 	int i, t, space;
1381 	unsigned sum;
1382 
1383 	sum = 1;
1384 	space = 0;
1385 	for (i = 0;;) {
1386 		switch (t = getc(f)) {
1387 		case '\0':
1388 			if ((flags & D_FORCEASCII) == 0)
1389 				return (RH_BINARY);
1390 			goto hashchar;
1391 		case '\r':
1392 			if (flags & D_STRIPCR) {
1393 				t = getc(f);
1394 				if (t == '\n')
1395 					break;
1396 				ungetc(t, f);
1397 			}
1398 			/* FALLTHROUGH */
1399 		case '\t':
1400 		case '\v':
1401 		case '\f':
1402 		case ' ':
1403 			if ((flags & (D_FOLDBLANKS|D_IGNOREBLANKS)) != 0) {
1404 				space++;
1405 				continue;
1406 			}
1407 			/* FALLTHROUGH */
1408 		default:
1409 		hashchar:
1410 			if (space && (flags & D_IGNOREBLANKS) == 0) {
1411 				i++;
1412 				space = 0;
1413 			}
1414 			sum = sum * 127 + chrtran(t);
1415 			i++;
1416 			continue;
1417 		case EOF:
1418 			if (i == 0)
1419 				return (RH_EOF);
1420 			/* FALLTHROUGH */
1421 		case '\n':
1422 			break;
1423 		}
1424 		break;
1425 	}
1426 	*hash = sum;
1427 	return (RH_OK);
1428 }
1429 
1430 static int
1431 asciifile(FILE *f)
1432 {
1433 	unsigned char buf[BUFSIZ];
1434 	size_t cnt;
1435 
1436 	if (f == NULL)
1437 		return (1);
1438 
1439 	rewind(f);
1440 	cnt = fread(buf, 1, sizeof(buf), f);
1441 	return (memchr(buf, '\0', cnt) == NULL);
1442 }
1443 
1444 #define begins_with(s, pre) (strncmp(s, pre, sizeof(pre) - 1) == 0)
1445 
1446 static char *
1447 match_function(const long *f, int pos, FILE *fp)
1448 {
1449 	unsigned char buf[FUNCTION_CONTEXT_SIZE];
1450 	size_t nc;
1451 	int last = lastline;
1452 	const char *state = NULL;
1453 
1454 	lastline = pos;
1455 	for (; pos > last; pos--) {
1456 		fseek(fp, f[pos - 1], SEEK_SET);
1457 		nc = f[pos] - f[pos - 1];
1458 		if (nc >= sizeof(buf))
1459 			nc = sizeof(buf) - 1;
1460 		nc = fread(buf, 1, nc, fp);
1461 		if (nc == 0)
1462 			continue;
1463 		buf[nc] = '\0';
1464 		buf[strcspn(buf, "\n")] = '\0';
1465 		if (most_recent_pat != NULL) {
1466 			int ret = regexec(&most_recent_re, buf, 0, NULL, 0);
1467 
1468 			if (ret != 0)
1469 				continue;
1470 			strlcpy(lastbuf, buf, sizeof(lastbuf));
1471 			lastmatchline = pos;
1472 			return (lastbuf);
1473 		} else if (isalpha(buf[0]) || buf[0] == '_' || buf[0] == '$'
1474 			|| buf[0] == '-' || buf[0] == '+') {
1475 			if (begins_with(buf, "private:")) {
1476 				if (!state)
1477 					state = " (private)";
1478 			} else if (begins_with(buf, "protected:")) {
1479 				if (!state)
1480 					state = " (protected)";
1481 			} else if (begins_with(buf, "public:")) {
1482 				if (!state)
1483 					state = " (public)";
1484 			} else {
1485 				strlcpy(lastbuf, buf, sizeof(lastbuf));
1486 				if (state)
1487 					strlcat(lastbuf, state, sizeof(lastbuf));
1488 				lastmatchline = pos;
1489 				return (lastbuf);
1490 			}
1491 		}
1492 	}
1493 	return (lastmatchline > 0 ? lastbuf : NULL);
1494 }
1495 
1496 /* dump accumulated "context" diff changes */
1497 static void
1498 dump_context_vec(FILE *f1, FILE *f2, int flags)
1499 {
1500 	struct context_vec *cvp = context_vec_start;
1501 	int lowa, upb, lowc, upd, do_output;
1502 	int a, b, c, d;
1503 	char ch, *f;
1504 
1505 	if (context_vec_start > context_vec_ptr)
1506 		return;
1507 
1508 	b = d = 0;		/* gcc */
1509 	lowa = MAX(1, cvp->a - diff_context);
1510 	upb = MIN((int)len[0], context_vec_ptr->b + diff_context);
1511 	lowc = MAX(1, cvp->c - diff_context);
1512 	upd = MIN((int)len[1], context_vec_ptr->d + diff_context);
1513 
1514 	printf("***************");
1515 	if (flags & (D_PROTOTYPE | D_MATCHLAST)) {
1516 		f = match_function(ixold, cvp->a - 1, f1);
1517 		if (f != NULL)
1518 			printf(" %s", f);
1519 	}
1520 	printf("\n*** ");
1521 	range(lowa, upb, ",");
1522 	printf(" ****\n");
1523 
1524 	/*
1525 	 * Output changes to the "old" file.  The first loop suppresses
1526 	 * output if there were no changes to the "old" file (we'll see
1527 	 * the "old" lines as context in the "new" list).
1528 	 */
1529 	do_output = 0;
1530 	for (; cvp <= context_vec_ptr; cvp++)
1531 		if (cvp->a <= cvp->b) {
1532 			cvp = context_vec_start;
1533 			do_output++;
1534 			break;
1535 		}
1536 	if (do_output) {
1537 		while (cvp <= context_vec_ptr) {
1538 			a = cvp->a;
1539 			b = cvp->b;
1540 			c = cvp->c;
1541 			d = cvp->d;
1542 
1543 			if (a <= b && c <= d)
1544 				ch = 'c';
1545 			else
1546 				ch = (a <= b) ? 'd' : 'a';
1547 
1548 			if (ch == 'a')
1549 				fetch(ixold, lowa, b, f1, ' ', 0, flags);
1550 			else {
1551 				fetch(ixold, lowa, a - 1, f1, ' ', 0, flags);
1552 				fetch(ixold, a, b, f1,
1553 				    ch == 'c' ? '!' : '-', 0, flags);
1554 			}
1555 			lowa = b + 1;
1556 			cvp++;
1557 		}
1558 		fetch(ixold, b + 1, upb, f1, ' ', 0, flags);
1559 	}
1560 	/* output changes to the "new" file */
1561 	printf("--- ");
1562 	range(lowc, upd, ",");
1563 	printf(" ----\n");
1564 
1565 	do_output = 0;
1566 	for (cvp = context_vec_start; cvp <= context_vec_ptr; cvp++)
1567 		if (cvp->c <= cvp->d) {
1568 			cvp = context_vec_start;
1569 			do_output++;
1570 			break;
1571 		}
1572 	if (do_output) {
1573 		while (cvp <= context_vec_ptr) {
1574 			a = cvp->a;
1575 			b = cvp->b;
1576 			c = cvp->c;
1577 			d = cvp->d;
1578 
1579 			if (a <= b && c <= d)
1580 				ch = 'c';
1581 			else
1582 				ch = (a <= b) ? 'd' : 'a';
1583 
1584 			if (ch == 'd')
1585 				fetch(ixnew, lowc, d, f2, ' ', 0, flags);
1586 			else {
1587 				fetch(ixnew, lowc, c - 1, f2, ' ', 0, flags);
1588 				fetch(ixnew, c, d, f2,
1589 				    ch == 'c' ? '!' : '+', 0, flags);
1590 			}
1591 			lowc = d + 1;
1592 			cvp++;
1593 		}
1594 		fetch(ixnew, d + 1, upd, f2, ' ', 0, flags);
1595 	}
1596 	context_vec_ptr = context_vec_start - 1;
1597 }
1598 
1599 /* dump accumulated "unified" diff changes */
1600 static void
1601 dump_unified_vec(FILE *f1, FILE *f2, int flags)
1602 {
1603 	struct context_vec *cvp = context_vec_start;
1604 	int lowa, upb, lowc, upd;
1605 	int a, b, c, d;
1606 	char ch, *f;
1607 
1608 	if (context_vec_start > context_vec_ptr)
1609 		return;
1610 
1611 	b = d = 0;		/* gcc */
1612 	lowa = MAX(1, cvp->a - diff_context);
1613 	upb = MIN((int)len[0], context_vec_ptr->b + diff_context);
1614 	lowc = MAX(1, cvp->c - diff_context);
1615 	upd = MIN((int)len[1], context_vec_ptr->d + diff_context);
1616 
1617 	printf("@@ -");
1618 	uni_range(lowa, upb);
1619 	printf(" +");
1620 	uni_range(lowc, upd);
1621 	printf(" @@");
1622 	if (flags & (D_PROTOTYPE | D_MATCHLAST)) {
1623 		f = match_function(ixold, cvp->a - 1, f1);
1624 		if (f != NULL)
1625 			printf(" %s", f);
1626 	}
1627 	printf("\n");
1628 
1629 	/*
1630 	 * Output changes in "unified" diff format--the old and new lines
1631 	 * are printed together.
1632 	 */
1633 	for (; cvp <= context_vec_ptr; cvp++) {
1634 		a = cvp->a;
1635 		b = cvp->b;
1636 		c = cvp->c;
1637 		d = cvp->d;
1638 
1639 		/*
1640 		 * c: both new and old changes
1641 		 * d: only changes in the old file
1642 		 * a: only changes in the new file
1643 		 */
1644 		if (a <= b && c <= d)
1645 			ch = 'c';
1646 		else
1647 			ch = (a <= b) ? 'd' : 'a';
1648 
1649 		switch (ch) {
1650 		case 'c':
1651 			fetch(ixold, lowa, a - 1, f1, ' ', 0, flags);
1652 			fetch(ixold, a, b, f1, '-', 0, flags);
1653 			fetch(ixnew, c, d, f2, '+', 0, flags);
1654 			break;
1655 		case 'd':
1656 			fetch(ixold, lowa, a - 1, f1, ' ', 0, flags);
1657 			fetch(ixold, a, b, f1, '-', 0, flags);
1658 			break;
1659 		case 'a':
1660 			fetch(ixnew, lowc, c - 1, f2, ' ', 0, flags);
1661 			fetch(ixnew, c, d, f2, '+', 0, flags);
1662 			break;
1663 		}
1664 		lowa = b + 1;
1665 		lowc = d + 1;
1666 	}
1667 	fetch(ixnew, d + 1, upd, f2, ' ', 0, flags);
1668 
1669 	context_vec_ptr = context_vec_start - 1;
1670 }
1671 
1672 static void
1673 print_header(const char *file1, const char *file2)
1674 {
1675 	const char *time_format;
1676 	char buf[256];
1677 	struct tm tm1, tm2, *tm_ptr1, *tm_ptr2;
1678 	int nsec1 = stb1.st_mtim.tv_nsec;
1679 	int nsec2 = stb2.st_mtim.tv_nsec;
1680 
1681 	time_format = "%Y-%m-%d %H:%M:%S";
1682 
1683 	if (cflag)
1684 		time_format = "%c";
1685 	tm_ptr1 = localtime_r(&stb1.st_mtime, &tm1);
1686 	tm_ptr2 = localtime_r(&stb2.st_mtime, &tm2);
1687 	if (label[0] != NULL)
1688 		printf("%s %s\n", diff_format == D_CONTEXT ? "***" : "---",
1689 		    label[0]);
1690 	else {
1691 		strftime(buf, sizeof(buf), time_format, tm_ptr1);
1692 		printf("%s %s\t%s", diff_format == D_CONTEXT ? "***" : "---",
1693 		    file1, buf);
1694 		if (!cflag) {
1695 			strftime(buf, sizeof(buf), "%z", tm_ptr1);
1696 			printf(".%.9d %s", nsec1, buf);
1697 		}
1698 		printf("\n");
1699 	}
1700 	if (label[1] != NULL)
1701 		printf("%s %s\n", diff_format == D_CONTEXT ? "---" : "+++",
1702 		    label[1]);
1703 	else {
1704 		strftime(buf, sizeof(buf), time_format, tm_ptr2);
1705 		printf("%s %s\t%s", diff_format == D_CONTEXT ? "---" : "+++",
1706 		    file2, buf);
1707 		if (!cflag) {
1708 			strftime(buf, sizeof(buf), "%z", tm_ptr2);
1709 			printf(".%.9d %s", nsec2, buf);
1710 		}
1711 		printf("\n");
1712 	}
1713 }
1714 
1715 /*
1716  * Prints n number of space characters either by using tab
1717  * or single space characters.
1718  * nc is the preceding number of characters
1719  */
1720 static void
1721 print_space(int nc, int n, int flags)
1722 {
1723 	int col, newcol, tabstop;
1724 
1725 	col = nc;
1726 	newcol = nc + n;
1727 	/* first, use tabs if allowed */
1728 	if ((flags & D_EXPANDTABS) == 0) {
1729 		while ((tabstop = roundup(col + 1, tabsize)) <= newcol) {
1730 			printf("\t");
1731 			col = tabstop;
1732 		}
1733 	}
1734 	/* finish with spaces */
1735 	printf("%*s", newcol - col, "");
1736 }
1737