1 /* $OpenBSD: diffreg.c,v 1.91 2016/03/01 20:57:35 natano Exp $ */
4 * SPDX-License-Identifier: BSD-4-Clause
6 * Copyright (C) Caldera International Inc. 2001-2002.
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
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
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.
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.
39 * Copyright (c) 1991, 1993
40 * The Regents of the University of California. All rights reserved.
42 * Redistribution and use in source and binary forms, with or without
43 * modification, are permitted provided that the following conditions
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.
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
66 * @(#)diffreg.c 8.1 (Berkeley) 6/6/93
69 #include <sys/cdefs.h>
70 __FBSDID("$FreeBSD$");
72 #include <sys/capsicum.h>
73 #include <sys/procdesc.h>
75 #include <sys/types.h>
76 #include <sys/event.h>
79 #include <capsicum_helpers.h>
98 #define _PATH_PR "/usr/bin/pr"
101 * diff - compare two files.
105 * Uses an algorithm due to Harold Stone, which finds
106 * a pair of longest identical subsequences in the two
109 * The major goal is to generate the match vector J.
110 * J[i] is the index of the line in file1 corresponding
111 * to line i file0. J[i] = 0 if there is no
112 * such line in file1.
114 * Lines are hashed so as to work in core. All potential
115 * matches are located by sorting the lines of each file
116 * on the hash (called ``value''). In particular, this
117 * collects the equivalence classes in file1 together.
118 * Subroutine equiv replaces the value of each line in
119 * file0 by the index of the first element of its
120 * matching equivalence in (the reordered) file1.
121 * To save space equiv squeezes file1 into a single
122 * array member in which the equivalence classes
123 * are simply concatenated, except that their first
124 * members are flagged by changing sign.
126 * Next the indices that point into member are unsorted into
127 * array class according to the original order of file0.
129 * The cleverness lies in routine stone. This marches
130 * through the lines of file0, developing a vector klist
131 * of "k-candidates". At step i a k-candidate is a matched
132 * pair of lines x,y (x in file0 y in file1) such that
133 * there is a common subsequence of length k
134 * between the first i lines of file0 and the first y
135 * lines of file1, but there is no such subsequence for
136 * any smaller y. x is the earliest possible mate to y
137 * that occurs in such a subsequence.
139 * Whenever any of the members of the equivalence class of
140 * lines in file1 matable to a line in file0 has serial number
141 * less than the y of some k-candidate, that k-candidate
142 * with the smallest such y is replaced. The new
143 * k-candidate is chained (via pred) to the current
144 * k-1 candidate so that the actual subsequence can
145 * be recovered. When a member has serial number greater
146 * that the y of all k-candidates, the klist is extended.
147 * At the end, the longest subsequence is pulled out
148 * and placed in the array J by unravel
150 * With J in hand, the matches there recorded are
151 * check'ed against reality to assure that no spurious
152 * matches have crept in due to hashing. If they have,
153 * they are broken, and "jackpot" is recorded--a harmless
154 * matter except that a true match for a spuriously
155 * mated line may now be unnecessarily reported as a change.
157 * Much of the complexity of the program comes simply
158 * from trying to minimize core utilization and
159 * maximize the range of doable problems by dynamically
160 * allocating what is needed and reusing what is not.
161 * The core requirements for problems larger than somewhat
162 * are (in words) 2*length(file0) + length(file1) +
163 * 3*(number of k-candidates installed), typically about
164 * 6n words for files of length n.
179 * The following struct is used to record change information when
180 * doing a "context" or "unified" diff. (see routine "change" to
181 * understand the highly mnemonic field names)
184 int a; /* start line in old file */
185 int b; /* end line in old file */
186 int c; /* start line in new file */
187 int d; /* end line in new file */
190 #define diff_output printf
191 static FILE *opentemp(const char *);
192 static void output(char *, FILE *, char *, FILE *, int);
193 static void check(FILE *, FILE *, int);
194 static void range(int, int, const char *);
195 static void uni_range(int, int);
196 static void dump_context_vec(FILE *, FILE *, int);
197 static void dump_unified_vec(FILE *, FILE *, int);
198 static void prepare(int, FILE *, size_t, int);
199 static void prune(void);
200 static void equiv(struct line *, int, struct line *, int, int *);
201 static void unravel(int);
202 static void unsort(struct line *, int, int *);
203 static void change(char *, FILE *, char *, FILE *, int, int, int, int, int *);
204 static void sort(struct line *, int);
205 static void print_header(const char *, const char *);
206 static int ignoreline(char *);
207 static int asciifile(FILE *);
208 static int fetch(long *, int, int, FILE *, int, int, int);
209 static int newcand(int, int, int);
210 static int search(int *, int, int);
211 static int skipline(FILE *);
212 static int isqrt(int);
213 static int stone(int *, int, int *, int *, int);
214 static int readhash(FILE *, int);
215 static int files_differ(FILE *, FILE *, int);
216 static char *match_function(const long *, int, FILE *);
217 static char *preadline(int, size_t, off_t);
219 static int *J; /* will be overlaid on class */
220 static int *class; /* will be overlaid on file[0] */
221 static int *klist; /* will be overlaid on file[0] after class */
222 static int *member; /* will be overlaid on file[1] */
224 static int inifdef; /* whether or not we are in a #ifdef block */
226 static int pref, suff; /* length of prefix and suffix */
228 static int anychange;
229 static long *ixnew; /* will be overlaid on file[1] */
230 static long *ixold; /* will be overlaid on klist */
231 static struct cand *clist; /* merely a free storage pot for candidates */
232 static int clistlen; /* the length of clist */
233 static struct line *sfile[2]; /* shortened by pruning common prefix/suffix */
234 static int (*chrtran)(int); /* translation table for case-folding */
235 static struct context_vec *context_vec_start;
236 static struct context_vec *context_vec_end;
237 static struct context_vec *context_vec_ptr;
239 #define FUNCTION_CONTEXT_SIZE 55
240 static char lastbuf[FUNCTION_CONTEXT_SIZE];
242 static int lastmatchline;
259 diffreg(char *file1, char *file2, int flags, int capsicum)
266 cap_rights_t rights_ro;
275 context_vec_ptr = context_vec_start - 1;
276 if (flags & D_IGNORECASE)
280 if (S_ISDIR(stb1.st_mode) != S_ISDIR(stb2.st_mode))
281 return (S_ISDIR(stb1.st_mode) ? D_MISMATCH1 : D_MISMATCH2);
282 if (strcmp(file1, "-") == 0 && strcmp(file2, "-") == 0)
285 if (flags & D_EMPTY1)
286 f1 = fopen(_PATH_DEVNULL, "r");
288 if (!S_ISREG(stb1.st_mode)) {
289 if ((f1 = opentemp(file1)) == NULL ||
290 fstat(fileno(f1), &stb1) < 0) {
295 } else if (strcmp(file1, "-") == 0)
298 f1 = fopen(file1, "r");
306 if (flags & D_EMPTY2)
307 f2 = fopen(_PATH_DEVNULL, "r");
309 if (!S_ISREG(stb2.st_mode)) {
310 if ((f2 = opentemp(file2)) == NULL ||
311 fstat(fileno(f2), &stb2) < 0) {
316 } else if (strcmp(file2, "-") == 0)
319 f2 = fopen(file2, "r");
328 /* redirect stdout to pr */
333 xasprintf(&header, "%s %s %s", diffargs, file1, file2);
334 signal(SIGPIPE, SIG_IGN);
338 switch ((pid = pdfork(&pr_pd, PD_CLOEXEC))) {
342 err(2, "No more processes");
345 if (pfd[0] != STDIN_FILENO) {
346 dup2(pfd[0], STDIN_FILENO);
350 execl(_PATH_PR, _PATH_PR, "-h", header, (char *)0);
355 if (pfd[1] != STDOUT_FILENO) {
356 ostdout = dup(STDOUT_FILENO);
357 dup2(pfd[1], STDOUT_FILENO);
366 e = xmalloc(sizeof(struct kevent));
367 EV_SET(e, pr_pd, EVFILT_PROCDESC, EV_ADD, NOTE_EXIT, 0,
369 if (kevent(kq, e, 1, NULL, 0, NULL) == -1)
375 cap_rights_init(&rights_ro, CAP_READ, CAP_FSTAT, CAP_SEEK);
376 if (cap_rights_limit(fileno(f1), &rights_ro) < 0
378 err(2, "unable to limit rights on: %s", file1);
379 if (cap_rights_limit(fileno(f2), &rights_ro) < 0 &&
381 err(2, "unable to limit rights on: %s", file2);
382 if (fileno(f1) == STDIN_FILENO || fileno(f2) == STDIN_FILENO) {
383 /* stding has already been limited */
384 if (caph_limit_stderr() == -1)
385 err(2, "unable to limit stderr");
386 if (caph_limit_stdout() == -1)
387 err(2, "unable to limit stdout");
388 } else if (caph_limit_stdio() == -1)
389 err(2, "unable to limit stdio");
391 caph_cache_catpages();
393 if (cap_enter() < 0 && errno != ENOSYS)
394 err(2, "unable to enter capability mode");
397 switch (files_differ(f1, f2, flags)) {
408 if ((flags & D_FORCEASCII) == 0 &&
409 (!asciifile(f1) || !asciifile(f2))) {
414 prepare(0, f1, stb1.st_size, flags);
415 prepare(1, f2, stb2.st_size, flags);
418 sort(sfile[0], slen[0]);
419 sort(sfile[1], slen[1]);
421 member = (int *)file[1];
422 equiv(sfile[0], slen[0], sfile[1], slen[1], member);
423 member = xreallocarray(member, slen[1] + 2, sizeof(*member));
425 class = (int *)file[0];
426 unsort(sfile[0], slen[0], class);
427 class = xreallocarray(class, slen[0] + 2, sizeof(*class));
429 klist = xcalloc(slen[0] + 2, sizeof(*klist));
432 clist = xcalloc(clistlen, sizeof(*clist));
433 i = stone(class, slen[0], member, klist, flags);
437 J = xreallocarray(J, len[0] + 2, sizeof(*J));
442 ixold = xreallocarray(ixold, len[0] + 2, sizeof(*ixold));
443 ixnew = xreallocarray(ixnew, len[1] + 2, sizeof(*ixnew));
444 check(f1, f2, flags);
445 output(file1, f1, file2, f2, flags);
446 if (ostdout != -1 && e != NULL) {
447 /* close the pipe to pr and restore stdout */
451 if (ostdout != STDOUT_FILENO) {
452 close(STDOUT_FILENO);
453 dup2(ostdout, STDOUT_FILENO);
456 if (kevent(kq, NULL, 0, e, 1, NULL) == -1)
460 if (WIFEXITED(wstatus) && WEXITSTATUS(wstatus) != 0)
461 errx(2, "pr exited abnormally");
462 else if (WIFSIGNALED(wstatus))
463 errx(2, "pr killed by signal %d",
482 * Check to see if the given files differ.
483 * Returns 0 if they are the same, 1 if different, and -1 on error.
484 * XXX - could use code from cmp(1) [faster]
487 files_differ(FILE *f1, FILE *f2, int flags)
489 char buf1[BUFSIZ], buf2[BUFSIZ];
492 if ((flags & (D_EMPTY1|D_EMPTY2)) || stb1.st_size != stb2.st_size ||
493 (stb1.st_mode & S_IFMT) != (stb2.st_mode & S_IFMT))
496 i = fread(buf1, 1, sizeof(buf1), f1);
497 j = fread(buf2, 1, sizeof(buf2), f2);
498 if ((!i && ferror(f1)) || (!j && ferror(f2)))
504 if (memcmp(buf1, buf2, i) != 0)
510 opentemp(const char *f)
512 char buf[BUFSIZ], tempfile[PATH_MAX];
516 if (strcmp(f, "-") == 0)
518 else if ((ifd = open(f, O_RDONLY, 0644)) < 0)
521 (void)strlcpy(tempfile, _PATH_TMP "/diff.XXXXXXXX", sizeof(tempfile));
523 if ((ofd = mkstemp(tempfile)) < 0) {
528 while ((nread = read(ifd, buf, BUFSIZ)) > 0) {
529 if (write(ofd, buf, nread) != nread) {
536 lseek(ofd, (off_t)0, SEEK_SET);
537 return (fdopen(ofd, "r"));
541 splice(char *dir, char *path)
546 dirlen = strlen(dir);
547 while (dirlen != 0 && dir[dirlen - 1] == '/')
549 if ((tail = strrchr(path, '/')) == NULL)
553 xasprintf(&buf, "%.*s/%s", (int)dirlen, dir, tail);
558 prepare(int i, FILE *fd, size_t filesize, int flags)
566 sz = MIN(filesize, SIZE_MAX) / 25;
570 p = xcalloc(sz + 3, sizeof(*p));
571 for (j = 0; (h = readhash(fd, flags));) {
574 p = xreallocarray(p, sz + 3, sizeof(*p));
587 for (pref = 0; pref < len[0] && pref < len[1] &&
588 file[0][pref + 1].value == file[1][pref + 1].value;
591 for (suff = 0; suff < len[0] - pref && suff < len[1] - pref &&
592 file[0][len[0] - suff].value == file[1][len[1] - suff].value;
595 for (j = 0; j < 2; j++) {
596 sfile[j] = file[j] + pref;
597 slen[j] = len[j] - pref - suff;
598 for (i = 0; i <= slen[j]; i++)
599 sfile[j][i].serial = i;
604 equiv(struct line *a, int n, struct line *b, int m, int *c)
609 while (i <= n && j <= m) {
610 if (a[i].value < b[j].value)
612 else if (a[i].value == b[j].value)
623 while (b[j + 1].value == b[j].value) {
631 /* Code taken from ping.c */
640 do { /* newton was a stinker */
645 } while ((x - y) > 1 || (x - y) < -1);
651 stone(int *a, int n, int *b, int *c, int flags)
654 int oldc, tc, oldl, sq;
655 u_int numtries, bound;
657 if (flags & D_MINIMAL)
661 bound = MAX(256, sq);
665 c[0] = newcand(0, 0, 0);
666 for (i = 1; i <= n; i++) {
675 if (y <= clist[oldc].y)
681 if (clist[c[l]].y <= y)
684 c[l] = newcand(i, y, oldc);
689 c[l] = newcand(i, y, oldc);
693 } while ((y = b[++j]) > 0 && numtries < bound);
699 newcand(int x, int y, int pred)
703 if (clen == clistlen) {
704 clistlen = clistlen * 11 / 10;
705 clist = xreallocarray(clist, clistlen, sizeof(*clist));
715 search(int *c, int k, int y)
719 if (clist[c[k]].y < y) /* quick look for typical case */
744 for (i = 0; i <= len[0]; i++)
745 J[i] = i <= pref ? i :
746 i > len[0] - suff ? i + len[1] - len[0] : 0;
747 for (q = clist + p; q->y != 0; q = clist + q->pred)
748 J[q->x + pref] = q->y + pref;
752 * Check does double duty:
753 * 1. ferret out any fortuitous correspondences due
754 * to confounding by hashing (which result in "jackpot")
755 * 2. collect random access indexes to the two files
758 check(FILE *f1, FILE *f2, int flags)
760 int i, j, jackpot, c, d;
766 ixold[0] = ixnew[0] = 0;
769 for (i = 1; i <= len[0]; i++) {
771 ixold[i] = ctold += skipline(f1);
775 ixnew[j] = ctnew += skipline(f2);
778 if (flags & (D_FOLDBLANKS|D_IGNOREBLANKS|D_IGNORECASE|D_STRIPCR)) {
783 * GNU diff ignores a missing newline
784 * in one file for -b or -w.
786 if (flags & (D_FOLDBLANKS|D_IGNOREBLANKS)) {
787 if (c == EOF && d == '\n') {
790 } else if (c == '\n' && d == EOF) {
797 if (flags & D_STRIPCR) {
799 if ((c = getc(f1)) == '\n') {
805 if ((d = getc(f2)) == '\n') {
811 if ((flags & D_FOLDBLANKS) && isspace(c) &&
817 } while (isspace(c = getc(f1)));
822 } while (isspace(d = getc(f2)));
823 } else if ((flags & D_IGNOREBLANKS)) {
824 while (isspace(c) && c != '\n') {
828 while (isspace(d) && d != '\n') {
833 if (chrtran(c) != chrtran(d)) {
836 if (c != '\n' && c != EOF)
837 ctold += skipline(f1);
838 if (d != '\n' && c != EOF)
839 ctnew += skipline(f2);
842 if (c == '\n' || c == EOF)
849 if ((c = getc(f1)) != (d = getc(f2))) {
852 if (c != '\n' && c != EOF)
853 ctold += skipline(f1);
854 if (d != '\n' && c != EOF)
855 ctnew += skipline(f2);
858 if (c == '\n' || c == EOF)
866 for (; j <= len[1]; j++) {
867 ixnew[j] = ctnew += skipline(f2);
871 * fprintf(stderr, "jackpot\n");
875 /* shellsort CACM #201 */
877 sort(struct line *a, int n)
879 struct line *ai, *aim, w;
884 for (j = 1; j <= n; j *= 2)
886 for (m /= 2; m != 0; m /= 2) {
888 for (j = 1; j <= k; j++) {
889 for (ai = &a[j]; ai > a; ai -= m) {
892 break; /* wraparound */
893 if (aim->value > ai[0].value ||
894 (aim->value == ai[0].value &&
895 aim->serial > ai[0].serial))
897 w.value = ai[0].value;
898 ai[0].value = aim->value;
899 aim->value = w.value;
900 w.serial = ai[0].serial;
901 ai[0].serial = aim->serial;
902 aim->serial = w.serial;
909 unsort(struct line *f, int l, int *b)
913 a = xcalloc(l + 1, sizeof(*a));
914 for (i = 1; i <= l; i++)
915 a[f[i].serial] = f[i].value;
916 for (i = 1; i <= l; i++)
926 for (i = 1; (c = getc(f)) != '\n' && c != EOF; i++)
932 output(char *file1, FILE *f1, char *file2, FILE *f2, int flags)
934 int m, i0, i1, j0, j1;
940 J[m + 1] = len[1] + 1;
941 if (diff_format != D_EDIT) {
942 for (i0 = 1; i0 <= m; i0 = i1 + 1) {
943 while (i0 <= m && J[i0] == J[i0 - 1] + 1)
947 while (i1 < m && J[i1 + 1] == 0)
951 change(file1, f1, file2, f2, i0, i1, j0, j1, &flags);
954 for (i0 = m; i0 >= 1; i0 = i1 - 1) {
955 while (i0 >= 1 && J[i0] == J[i0 + 1] - 1 && J[i0] != 0)
959 while (i1 > 1 && J[i1 - 1] == 0)
963 change(file1, f1, file2, f2, i1, i0, j1, j0, &flags);
967 change(file1, f1, file2, f2, 1, 0, 1, len[1], &flags);
968 if (diff_format == D_IFDEF || diff_format == D_GFORMAT) {
971 if ((c = getc(f1)) == EOF)
973 diff_output("%c", c);
977 if (anychange != 0) {
978 if (diff_format == D_CONTEXT)
979 dump_context_vec(f1, f2, flags);
980 else if (diff_format == D_UNIFIED)
981 dump_unified_vec(f1, f2, flags);
986 range(int a, int b, const char *separator)
988 diff_output("%d", a > b ? b : a);
990 diff_output("%s%d", separator, b);
994 uni_range(int a, int b)
997 diff_output("%d,%d", a, b - a + 1);
999 diff_output("%d", b);
1001 diff_output("%d,0", b);
1005 preadline(int fd, size_t rlen, off_t off)
1010 line = xmalloc(rlen + 1);
1011 if ((nr = pread(fd, line, rlen, off)) < 0)
1012 err(2, "preadline");
1013 if (nr > 0 && line[nr-1] == '\n')
1020 ignoreline(char *line)
1024 ret = regexec(&ignore_re, line, 0, NULL, 0);
1026 return (ret == 0); /* if it matched, it should be ignored. */
1030 * Indicate that there is a difference between lines a and b of the from file
1031 * to get to lines c to d of the to file. If a is greater then b then there
1032 * are no lines in the from file involved and this means that there were
1033 * lines appended (beginning at b). If c is greater than d then there are
1034 * lines missing from the to file.
1037 change(char *file1, FILE *f1, char *file2, FILE *f2, int a, int b, int c, int d,
1040 static size_t max_context = 64;
1046 if ((diff_format != D_IFDEF || diff_format == D_GFORMAT) &&
1049 if (ignore_pats != NULL) {
1052 * All lines in the change, insert, or delete must
1053 * match an ignore pattern for the change to be
1056 if (a <= b) { /* Changes and deletes. */
1057 for (i = a; i <= b; i++) {
1058 line = preadline(fileno(f1),
1059 ixold[i] - ixold[i - 1], ixold[i - 1]);
1060 if (!ignoreline(line))
1064 if (a > b || c <= d) { /* Changes and inserts. */
1065 for (i = c; i <= d; i++) {
1066 line = preadline(fileno(f2),
1067 ixnew[i] - ixnew[i - 1], ixnew[i - 1]);
1068 if (!ignoreline(line))
1075 if (*pflags & D_HEADER && diff_format != D_BRIEF) {
1076 diff_output("%s %s %s\n", diffargs, file1, file2);
1077 *pflags &= ~D_HEADER;
1079 if (diff_format == D_CONTEXT || diff_format == D_UNIFIED) {
1081 * Allocate change records as needed.
1083 if (context_vec_ptr == context_vec_end - 1) {
1084 ptrdiff_t offset = context_vec_ptr - context_vec_start;
1086 context_vec_start = xreallocarray(context_vec_start,
1087 max_context, sizeof(*context_vec_start));
1088 context_vec_end = context_vec_start + max_context;
1089 context_vec_ptr = context_vec_start + offset;
1091 if (anychange == 0) {
1093 * Print the context/unidiff header first time through.
1095 print_header(file1, file2);
1097 } else if (a > context_vec_ptr->b + (2 * diff_context) + 1 &&
1098 c > context_vec_ptr->d + (2 * diff_context) + 1) {
1100 * If this change is more than 'diff_context' lines from the
1101 * previous change, dump the record and reset it.
1103 if (diff_format == D_CONTEXT)
1104 dump_context_vec(f1, f2, *pflags);
1106 dump_unified_vec(f1, f2, *pflags);
1109 context_vec_ptr->a = a;
1110 context_vec_ptr->b = b;
1111 context_vec_ptr->c = c;
1112 context_vec_ptr->d = d;
1117 switch (diff_format) {
1123 diff_output("%c", a > b ? 'a' : c > d ? 'd' : 'c');
1124 if (diff_format == D_NORMAL)
1129 diff_output("%c", a > b ? 'a' : c > d ? 'd' : 'c');
1135 diff_output("a%d %d\n", b, d - c + 1);
1137 diff_output("d%d %d\n", a, b - a + 1);
1139 /* add changed lines */
1140 diff_output("a%d %d\n", b, d - c + 1);
1144 if (diff_format == D_GFORMAT) {
1146 /* print through if append (a>b), else to (nb: 0 vs 1 orig) */
1147 nc = ixold[a > b ? b : a - 1] - curpos;
1148 for (i = 0; i < nc; i++)
1149 diff_output("%c", getc(f1));
1150 for (walk = group_format; *walk != '\0'; walk++) {
1155 fetch(ixold, a, b, f1, '<', 1, *pflags);
1158 fetch(ixnew, c, d, f2, '>', 0, *pflags);
1161 diff_output("%%%c", *walk);
1166 diff_output("%c", *walk);
1169 if (diff_format == D_NORMAL || diff_format == D_IFDEF) {
1170 fetch(ixold, a, b, f1, '<', 1, *pflags);
1171 if (a <= b && c <= d && diff_format == D_NORMAL)
1172 diff_output("---\n");
1175 if (diff_format != D_GFORMAT)
1176 f = fetch(ixnew, c, d, f2, diff_format == D_NORMAL ? '>' : '\0', 0, *pflags);
1177 if (f != 0 && diff_format == D_EDIT) {
1179 * A non-zero return value for D_EDIT indicates that the
1180 * last line printed was a bare dot (".") that has been
1181 * escaped as ".." to prevent ed(1) from misinterpreting
1182 * it. We have to add a substitute command to change this
1183 * back and restart where we left off.
1186 diff_output("%ds/.//\n", a + f - 1);
1192 if ((diff_format == D_EDIT || diff_format == D_REVERSE) && c <= d)
1195 diff_output("#endif /* %s */\n", ifdefname);
1201 fetch(long *f, int a, int b, FILE *lb, int ch, int oldfile, int flags)
1203 int i, j, c, lastc, col, nc;
1207 * When doing #ifdef's, copy down to current line
1208 * if this is the first file, so that stuff makes it to output.
1210 if ((diff_format == D_IFDEF) && oldfile) {
1211 long curpos = ftell(lb);
1212 /* print through if append (a>b), else to (nb: 0 vs 1 orig) */
1213 nc = f[a > b ? b : a - 1] - curpos;
1214 for (i = 0; i < nc; i++)
1215 diff_output("%c", getc(lb));
1219 if (diff_format == D_IFDEF) {
1221 diff_output("#else /* %s%s */\n",
1222 oldfile == 1 ? "!" : "", ifdefname);
1225 diff_output("#ifndef %s\n", ifdefname);
1227 diff_output("#ifdef %s\n", ifdefname);
1229 inifdef = 1 + oldfile;
1231 for (i = a; i <= b; i++) {
1232 fseek(lb, f[i - 1], SEEK_SET);
1233 nc = f[i] - f[i - 1];
1234 if ((diff_format != D_IFDEF && diff_format != D_GFORMAT) &&
1236 diff_output("%c", ch);
1237 if (Tflag && (diff_format == D_NORMAL || diff_format == D_CONTEXT
1238 || diff_format == D_UNIFIED))
1240 else if (diff_format != D_UNIFIED)
1244 for (j = 0, lastc = '\0'; j < nc; j++, lastc = c) {
1245 if ((c = getc(lb)) == EOF) {
1246 if (diff_format == D_EDIT || diff_format == D_REVERSE ||
1247 diff_format == D_NREVERSE)
1248 warnx("No newline at end of file");
1250 diff_output("\n\\ No newline at end of "
1254 if (c == '\t' && (flags & D_EXPANDTABS)) {
1255 newcol = ((col/tabsize)+1)*tabsize;
1258 } while (++col < newcol);
1260 if (diff_format == D_EDIT && j == 1 && c == '\n'
1263 * Don't print a bare "." line
1264 * since that will confuse ed(1).
1265 * Print ".." instead and return,
1266 * giving the caller an offset
1267 * from which to restart.
1272 diff_output("%c", c);
1281 * Hash function taken from Robert Sedgewick, Algorithms in C, 3d ed., p 578.
1284 readhash(FILE *f, int flags)
1291 if ((flags & (D_FOLDBLANKS|D_IGNOREBLANKS)) == 0) {
1292 if (flags & D_IGNORECASE)
1293 for (i = 0; (t = getc(f)) != '\n'; i++) {
1294 if (flags & D_STRIPCR && t == '\r') {
1305 sum = sum * 127 + chrtran(t);
1308 for (i = 0; (t = getc(f)) != '\n'; i++) {
1309 if (flags & D_STRIPCR && t == '\r') {
1320 sum = sum * 127 + t;
1324 switch (t = getc(f)) {
1333 if (space && (flags & D_IGNOREBLANKS) == 0) {
1337 sum = sum * 127 + chrtran(t);
1351 * There is a remote possibility that we end up with a zero sum.
1352 * Zero is used as an EOF marker, so return 1 instead.
1354 return (sum == 0 ? 1 : sum);
1360 unsigned char buf[BUFSIZ];
1367 cnt = fread(buf, 1, sizeof(buf), f);
1368 return (memchr(buf, '\0', cnt) == NULL);
1371 #define begins_with(s, pre) (strncmp(s, pre, sizeof(pre)-1) == 0)
1374 match_function(const long *f, int pos, FILE *fp)
1376 unsigned char buf[FUNCTION_CONTEXT_SIZE];
1378 int last = lastline;
1379 const char *state = NULL;
1382 while (pos > last) {
1383 fseek(fp, f[pos - 1], SEEK_SET);
1384 nc = f[pos] - f[pos - 1];
1385 if (nc >= sizeof(buf))
1386 nc = sizeof(buf) - 1;
1387 nc = fread(buf, 1, nc, fp);
1390 buf[strcspn(buf, "\n")] = '\0';
1391 if (isalpha(buf[0]) || buf[0] == '_' || buf[0] == '$') {
1392 if (begins_with(buf, "private:")) {
1394 state = " (private)";
1395 } else if (begins_with(buf, "protected:")) {
1397 state = " (protected)";
1398 } else if (begins_with(buf, "public:")) {
1400 state = " (public)";
1402 strlcpy(lastbuf, buf, sizeof lastbuf);
1404 strlcat(lastbuf, state,
1406 lastmatchline = pos;
1413 return lastmatchline > 0 ? lastbuf : NULL;
1416 /* dump accumulated "context" diff changes */
1418 dump_context_vec(FILE *f1, FILE *f2, int flags)
1420 struct context_vec *cvp = context_vec_start;
1421 int lowa, upb, lowc, upd, do_output;
1425 if (context_vec_start > context_vec_ptr)
1428 b = d = 0; /* gcc */
1429 lowa = MAX(1, cvp->a - diff_context);
1430 upb = MIN(len[0], context_vec_ptr->b + diff_context);
1431 lowc = MAX(1, cvp->c - diff_context);
1432 upd = MIN(len[1], context_vec_ptr->d + diff_context);
1434 diff_output("***************");
1435 if ((flags & D_PROTOTYPE)) {
1436 f = match_function(ixold, lowa-1, f1);
1438 diff_output(" %s", f);
1440 diff_output("\n*** ");
1441 range(lowa, upb, ",");
1442 diff_output(" ****\n");
1445 * Output changes to the "old" file. The first loop suppresses
1446 * output if there were no changes to the "old" file (we'll see
1447 * the "old" lines as context in the "new" list).
1450 for (; cvp <= context_vec_ptr; cvp++)
1451 if (cvp->a <= cvp->b) {
1452 cvp = context_vec_start;
1457 while (cvp <= context_vec_ptr) {
1463 if (a <= b && c <= d)
1466 ch = (a <= b) ? 'd' : 'a';
1469 fetch(ixold, lowa, b, f1, ' ', 0, flags);
1471 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags);
1472 fetch(ixold, a, b, f1,
1473 ch == 'c' ? '!' : '-', 0, flags);
1478 fetch(ixold, b + 1, upb, f1, ' ', 0, flags);
1480 /* output changes to the "new" file */
1481 diff_output("--- ");
1482 range(lowc, upd, ",");
1483 diff_output(" ----\n");
1486 for (cvp = context_vec_start; cvp <= context_vec_ptr; cvp++)
1487 if (cvp->c <= cvp->d) {
1488 cvp = context_vec_start;
1493 while (cvp <= context_vec_ptr) {
1499 if (a <= b && c <= d)
1502 ch = (a <= b) ? 'd' : 'a';
1505 fetch(ixnew, lowc, d, f2, ' ', 0, flags);
1507 fetch(ixnew, lowc, c - 1, f2, ' ', 0, flags);
1508 fetch(ixnew, c, d, f2,
1509 ch == 'c' ? '!' : '+', 0, flags);
1514 fetch(ixnew, d + 1, upd, f2, ' ', 0, flags);
1516 context_vec_ptr = context_vec_start - 1;
1519 /* dump accumulated "unified" diff changes */
1521 dump_unified_vec(FILE *f1, FILE *f2, int flags)
1523 struct context_vec *cvp = context_vec_start;
1524 int lowa, upb, lowc, upd;
1528 if (context_vec_start > context_vec_ptr)
1531 b = d = 0; /* gcc */
1532 lowa = MAX(1, cvp->a - diff_context);
1533 upb = MIN(len[0], context_vec_ptr->b + diff_context);
1534 lowc = MAX(1, cvp->c - diff_context);
1535 upd = MIN(len[1], context_vec_ptr->d + diff_context);
1537 diff_output("@@ -");
1538 uni_range(lowa, upb);
1540 uni_range(lowc, upd);
1542 if ((flags & D_PROTOTYPE)) {
1543 f = match_function(ixold, lowa-1, f1);
1545 diff_output(" %s", f);
1550 * Output changes in "unified" diff format--the old and new lines
1551 * are printed together.
1553 for (; cvp <= context_vec_ptr; cvp++) {
1560 * c: both new and old changes
1561 * d: only changes in the old file
1562 * a: only changes in the new file
1564 if (a <= b && c <= d)
1567 ch = (a <= b) ? 'd' : 'a';
1571 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags);
1572 fetch(ixold, a, b, f1, '-', 0, flags);
1573 fetch(ixnew, c, d, f2, '+', 0, flags);
1576 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags);
1577 fetch(ixold, a, b, f1, '-', 0, flags);
1580 fetch(ixnew, lowc, c - 1, f2, ' ', 0, flags);
1581 fetch(ixnew, c, d, f2, '+', 0, flags);
1587 fetch(ixnew, d + 1, upd, f2, ' ', 0, flags);
1589 context_vec_ptr = context_vec_start - 1;
1593 print_header(const char *file1, const char *file2)
1595 const char *time_format;
1600 struct tm tm1, tm2, *tm_ptr1, *tm_ptr2;
1601 int nsec1 = stb1.st_mtim.tv_nsec;
1602 int nsec2 = stb2.st_mtim.tv_nsec;
1604 time_format = "%Y-%m-%d %H:%M:%S";
1608 tm_ptr1 = localtime_r(&stb1.st_mtime, &tm1);
1609 tm_ptr2 = localtime_r(&stb2.st_mtime, &tm2);
1610 strftime(buf1, 256, time_format, tm_ptr1);
1611 strftime(buf2, 256, time_format, tm_ptr2);
1613 strftime(end1, 10, "%z", tm_ptr1);
1614 strftime(end2, 10, "%z", tm_ptr2);
1615 sprintf(buf1, "%s.%.9d %s", buf1, nsec1, end1);
1616 sprintf(buf2, "%s.%.9d %s", buf2, nsec2, end2);
1618 if (label[0] != NULL)
1619 diff_output("%s %s\n", diff_format == D_CONTEXT ? "***" : "---",
1622 diff_output("%s %s\t%s\n", diff_format == D_CONTEXT ? "***" : "---",
1624 if (label[1] != NULL)
1625 diff_output("%s %s\n", diff_format == D_CONTEXT ? "---" : "+++",
1628 diff_output("%s %s\t%s\n", diff_format == D_CONTEXT ? "---" : "+++",