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>
75 #include <capsicum_helpers.h>
94 * diff - compare two files.
98 * Uses an algorithm due to Harold Stone, which finds
99 * a pair of longest identical subsequences in the two
102 * The major goal is to generate the match vector J.
103 * J[i] is the index of the line in file1 corresponding
104 * to line i file0. J[i] = 0 if there is no
105 * such line in file1.
107 * Lines are hashed so as to work in core. All potential
108 * matches are located by sorting the lines of each file
109 * on the hash (called ``value''). In particular, this
110 * collects the equivalence classes in file1 together.
111 * Subroutine equiv replaces the value of each line in
112 * file0 by the index of the first element of its
113 * matching equivalence in (the reordered) file1.
114 * To save space equiv squeezes file1 into a single
115 * array member in which the equivalence classes
116 * are simply concatenated, except that their first
117 * members are flagged by changing sign.
119 * Next the indices that point into member are unsorted into
120 * array class according to the original order of file0.
122 * The cleverness lies in routine stone. This marches
123 * through the lines of file0, developing a vector klist
124 * of "k-candidates". At step i a k-candidate is a matched
125 * pair of lines x,y (x in file0 y in file1) such that
126 * there is a common subsequence of length k
127 * between the first i lines of file0 and the first y
128 * lines of file1, but there is no such subsequence for
129 * any smaller y. x is the earliest possible mate to y
130 * that occurs in such a subsequence.
132 * Whenever any of the members of the equivalence class of
133 * lines in file1 matable to a line in file0 has serial number
134 * less than the y of some k-candidate, that k-candidate
135 * with the smallest such y is replaced. The new
136 * k-candidate is chained (via pred) to the current
137 * k-1 candidate so that the actual subsequence can
138 * be recovered. When a member has serial number greater
139 * that the y of all k-candidates, the klist is extended.
140 * At the end, the longest subsequence is pulled out
141 * and placed in the array J by unravel
143 * With J in hand, the matches there recorded are
144 * check'ed against reality to assure that no spurious
145 * matches have crept in due to hashing. If they have,
146 * they are broken, and "jackpot" is recorded--a harmless
147 * matter except that a true match for a spuriously
148 * mated line may now be unnecessarily reported as a change.
150 * Much of the complexity of the program comes simply
151 * from trying to minimize core utilization and
152 * maximize the range of doable problems by dynamically
153 * allocating what is needed and reusing what is not.
154 * The core requirements for problems larger than somewhat
155 * are (in words) 2*length(file0) + length(file1) +
156 * 3*(number of k-candidates installed), typically about
157 * 6n words for files of length n.
172 * The following struct is used to record change information when
173 * doing a "context" or "unified" diff. (see routine "change" to
174 * understand the highly mnemonic field names)
177 int a; /* start line in old file */
178 int b; /* end line in old file */
179 int c; /* start line in new file */
180 int d; /* end line in new file */
183 #define diff_output printf
184 static FILE *opentemp(const char *);
185 static void output(char *, FILE *, char *, FILE *, int);
186 static void check(FILE *, FILE *, int);
187 static void range(int, int, const char *);
188 static void uni_range(int, int);
189 static void dump_context_vec(FILE *, FILE *, int);
190 static void dump_unified_vec(FILE *, FILE *, int);
191 static void prepare(int, FILE *, size_t, int);
192 static void prune(void);
193 static void equiv(struct line *, int, struct line *, int, int *);
194 static void unravel(int);
195 static void unsort(struct line *, int, int *);
196 static void change(char *, FILE *, char *, FILE *, int, int, int, int, int *);
197 static void sort(struct line *, int);
198 static void print_header(const char *, const char *);
199 static bool ignoreline_pattern(char *);
200 static bool ignoreline(char *, bool);
201 static int asciifile(FILE *);
202 static int fetch(long *, int, int, FILE *, int, int, int);
203 static int newcand(int, int, int);
204 static int search(int *, int, int);
205 static int skipline(FILE *);
206 static int isqrt(int);
207 static int stone(int *, int, int *, int *, int);
208 static int readhash(FILE *, int);
209 static int files_differ(FILE *, FILE *, int);
210 static char *match_function(const long *, int, FILE *);
211 static char *preadline(int, size_t, off_t);
213 static int *J; /* will be overlaid on class */
214 static int *class; /* will be overlaid on file[0] */
215 static int *klist; /* will be overlaid on file[0] after class */
216 static int *member; /* will be overlaid on file[1] */
218 static int inifdef; /* whether or not we are in a #ifdef block */
220 static int pref, suff; /* length of prefix and suffix */
222 static int anychange;
223 static long *ixnew; /* will be overlaid on file[1] */
224 static long *ixold; /* will be overlaid on klist */
225 static struct cand *clist; /* merely a free storage pot for candidates */
226 static int clistlen; /* the length of clist */
227 static struct line *sfile[2]; /* shortened by pruning common prefix/suffix */
228 static int (*chrtran)(int); /* translation table for case-folding */
229 static struct context_vec *context_vec_start;
230 static struct context_vec *context_vec_end;
231 static struct context_vec *context_vec_ptr;
233 #define FUNCTION_CONTEXT_SIZE 55
234 static char lastbuf[FUNCTION_CONTEXT_SIZE];
236 static int lastmatchline;
253 diffreg(char *file1, char *file2, int flags, int capsicum)
257 struct pr *pr = NULL;
258 cap_rights_t rights_ro;
265 context_vec_ptr = context_vec_start - 1;
266 if (flags & D_IGNORECASE)
270 if (S_ISDIR(stb1.st_mode) != S_ISDIR(stb2.st_mode))
271 return (S_ISDIR(stb1.st_mode) ? D_MISMATCH1 : D_MISMATCH2);
272 if (strcmp(file1, "-") == 0 && strcmp(file2, "-") == 0)
275 if (flags & D_EMPTY1)
276 f1 = fopen(_PATH_DEVNULL, "r");
278 if (!S_ISREG(stb1.st_mode)) {
279 if ((f1 = opentemp(file1)) == NULL ||
280 fstat(fileno(f1), &stb1) < 0) {
285 } else if (strcmp(file1, "-") == 0)
288 f1 = fopen(file1, "r");
296 if (flags & D_EMPTY2)
297 f2 = fopen(_PATH_DEVNULL, "r");
299 if (!S_ISREG(stb2.st_mode)) {
300 if ((f2 = opentemp(file2)) == NULL ||
301 fstat(fileno(f2), &stb2) < 0) {
306 } else if (strcmp(file2, "-") == 0)
309 f2 = fopen(file2, "r");
318 pr = start_pr(file1, file2);
321 cap_rights_init(&rights_ro, CAP_READ, CAP_FSTAT, CAP_SEEK);
322 if (caph_rights_limit(fileno(f1), &rights_ro) < 0)
323 err(2, "unable to limit rights on: %s", file1);
324 if (caph_rights_limit(fileno(f2), &rights_ro) < 0)
325 err(2, "unable to limit rights on: %s", file2);
326 if (fileno(f1) == STDIN_FILENO || fileno(f2) == STDIN_FILENO) {
327 /* stding has already been limited */
328 if (caph_limit_stderr() == -1)
329 err(2, "unable to limit stderr");
330 if (caph_limit_stdout() == -1)
331 err(2, "unable to limit stdout");
332 } else if (caph_limit_stdio() == -1)
333 err(2, "unable to limit stdio");
335 caph_cache_catpages();
337 if (caph_enter() < 0)
338 err(2, "unable to enter capability mode");
341 switch (files_differ(f1, f2, flags)) {
352 if ((flags & D_FORCEASCII) == 0 &&
353 (!asciifile(f1) || !asciifile(f2))) {
358 prepare(0, f1, stb1.st_size, flags);
359 prepare(1, f2, stb2.st_size, flags);
362 sort(sfile[0], slen[0]);
363 sort(sfile[1], slen[1]);
365 member = (int *)file[1];
366 equiv(sfile[0], slen[0], sfile[1], slen[1], member);
367 member = xreallocarray(member, slen[1] + 2, sizeof(*member));
369 class = (int *)file[0];
370 unsort(sfile[0], slen[0], class);
371 class = xreallocarray(class, slen[0] + 2, sizeof(*class));
373 klist = xcalloc(slen[0] + 2, sizeof(*klist));
376 clist = xcalloc(clistlen, sizeof(*clist));
377 i = stone(class, slen[0], member, klist, flags);
381 J = xreallocarray(J, len[0] + 2, sizeof(*J));
386 ixold = xreallocarray(ixold, len[0] + 2, sizeof(*ixold));
387 ixnew = xreallocarray(ixnew, len[1] + 2, sizeof(*ixnew));
388 check(f1, f2, flags);
389 output(file1, f1, file2, f2, flags);
408 * Check to see if the given files differ.
409 * Returns 0 if they are the same, 1 if different, and -1 on error.
410 * XXX - could use code from cmp(1) [faster]
413 files_differ(FILE *f1, FILE *f2, int flags)
415 char buf1[BUFSIZ], buf2[BUFSIZ];
418 if ((flags & (D_EMPTY1|D_EMPTY2)) || stb1.st_size != stb2.st_size ||
419 (stb1.st_mode & S_IFMT) != (stb2.st_mode & S_IFMT))
422 i = fread(buf1, 1, sizeof(buf1), f1);
423 j = fread(buf2, 1, sizeof(buf2), f2);
424 if ((!i && ferror(f1)) || (!j && ferror(f2)))
430 if (memcmp(buf1, buf2, i) != 0)
436 opentemp(const char *f)
438 char buf[BUFSIZ], tempfile[PATH_MAX];
442 if (strcmp(f, "-") == 0)
444 else if ((ifd = open(f, O_RDONLY, 0644)) < 0)
447 (void)strlcpy(tempfile, _PATH_TMP "/diff.XXXXXXXX", sizeof(tempfile));
449 if ((ofd = mkstemp(tempfile)) < 0) {
454 while ((nread = read(ifd, buf, BUFSIZ)) > 0) {
455 if (write(ofd, buf, nread) != nread) {
462 lseek(ofd, (off_t)0, SEEK_SET);
463 return (fdopen(ofd, "r"));
467 splice(char *dir, char *path)
472 dirlen = strlen(dir);
473 while (dirlen != 0 && dir[dirlen - 1] == '/')
475 if ((tail = strrchr(path, '/')) == NULL)
479 xasprintf(&buf, "%.*s/%s", (int)dirlen, dir, tail);
484 prepare(int i, FILE *fd, size_t filesize, int flags)
492 sz = MIN(filesize, SIZE_MAX) / 25;
496 p = xcalloc(sz + 3, sizeof(*p));
497 for (j = 0; (h = readhash(fd, flags));) {
500 p = xreallocarray(p, sz + 3, sizeof(*p));
513 for (pref = 0; pref < len[0] && pref < len[1] &&
514 file[0][pref + 1].value == file[1][pref + 1].value;
517 for (suff = 0; suff < len[0] - pref && suff < len[1] - pref &&
518 file[0][len[0] - suff].value == file[1][len[1] - suff].value;
521 for (j = 0; j < 2; j++) {
522 sfile[j] = file[j] + pref;
523 slen[j] = len[j] - pref - suff;
524 for (i = 0; i <= slen[j]; i++)
525 sfile[j][i].serial = i;
530 equiv(struct line *a, int n, struct line *b, int m, int *c)
535 while (i <= n && j <= m) {
536 if (a[i].value < b[j].value)
538 else if (a[i].value == b[j].value)
549 while (b[j + 1].value == b[j].value) {
557 /* Code taken from ping.c */
566 do { /* newton was a stinker */
571 } while ((x - y) > 1 || (x - y) < -1);
577 stone(int *a, int n, int *b, int *c, int flags)
580 int oldc, tc, oldl, sq;
581 u_int numtries, bound;
583 if (flags & D_MINIMAL)
587 bound = MAX(256, sq);
591 c[0] = newcand(0, 0, 0);
592 for (i = 1; i <= n; i++) {
601 if (y <= clist[oldc].y)
607 if (clist[c[l]].y <= y)
610 c[l] = newcand(i, y, oldc);
615 c[l] = newcand(i, y, oldc);
619 } while ((y = b[++j]) > 0 && numtries < bound);
625 newcand(int x, int y, int pred)
629 if (clen == clistlen) {
630 clistlen = clistlen * 11 / 10;
631 clist = xreallocarray(clist, clistlen, sizeof(*clist));
641 search(int *c, int k, int y)
645 if (clist[c[k]].y < y) /* quick look for typical case */
670 for (i = 0; i <= len[0]; i++)
671 J[i] = i <= pref ? i :
672 i > len[0] - suff ? i + len[1] - len[0] : 0;
673 for (q = clist + p; q->y != 0; q = clist + q->pred)
674 J[q->x + pref] = q->y + pref;
678 * Check does double duty:
679 * 1. ferret out any fortuitous correspondences due
680 * to confounding by hashing (which result in "jackpot")
681 * 2. collect random access indexes to the two files
684 check(FILE *f1, FILE *f2, int flags)
686 int i, j, jackpot, c, d;
692 ixold[0] = ixnew[0] = 0;
695 for (i = 1; i <= len[0]; i++) {
697 ixold[i] = ctold += skipline(f1);
701 ixnew[j] = ctnew += skipline(f2);
704 if (flags & (D_FOLDBLANKS|D_IGNOREBLANKS|D_IGNORECASE|D_STRIPCR)) {
709 * GNU diff ignores a missing newline
710 * in one file for -b or -w.
712 if (flags & (D_FOLDBLANKS|D_IGNOREBLANKS)) {
713 if (c == EOF && d == '\n') {
716 } else if (c == '\n' && d == EOF) {
723 if (flags & D_STRIPCR && (c == '\r' || d == '\r')) {
725 if ((c = getc(f1)) == '\n') {
732 if ((d = getc(f2)) == '\n') {
740 if ((flags & D_FOLDBLANKS) && isspace(c) &&
746 } while (isspace(c = getc(f1)));
751 } while (isspace(d = getc(f2)));
752 } else if ((flags & D_IGNOREBLANKS)) {
753 while (isspace(c) && c != '\n') {
757 while (isspace(d) && d != '\n') {
762 if (chrtran(c) != chrtran(d)) {
765 if (c != '\n' && c != EOF)
766 ctold += skipline(f1);
767 if (d != '\n' && c != EOF)
768 ctnew += skipline(f2);
771 if (c == '\n' || c == EOF)
778 if ((c = getc(f1)) != (d = getc(f2))) {
781 if (c != '\n' && c != EOF)
782 ctold += skipline(f1);
783 if (d != '\n' && c != EOF)
784 ctnew += skipline(f2);
787 if (c == '\n' || c == EOF)
795 for (; j <= len[1]; j++) {
796 ixnew[j] = ctnew += skipline(f2);
800 * fprintf(stderr, "jackpot\n");
804 /* shellsort CACM #201 */
806 sort(struct line *a, int n)
808 struct line *ai, *aim, w;
813 for (j = 1; j <= n; j *= 2)
815 for (m /= 2; m != 0; m /= 2) {
817 for (j = 1; j <= k; j++) {
818 for (ai = &a[j]; ai > a; ai -= m) {
821 break; /* wraparound */
822 if (aim->value > ai[0].value ||
823 (aim->value == ai[0].value &&
824 aim->serial > ai[0].serial))
826 w.value = ai[0].value;
827 ai[0].value = aim->value;
828 aim->value = w.value;
829 w.serial = ai[0].serial;
830 ai[0].serial = aim->serial;
831 aim->serial = w.serial;
838 unsort(struct line *f, int l, int *b)
842 a = xcalloc(l + 1, sizeof(*a));
843 for (i = 1; i <= l; i++)
844 a[f[i].serial] = f[i].value;
845 for (i = 1; i <= l; i++)
855 for (i = 1; (c = getc(f)) != '\n' && c != EOF; i++)
861 output(char *file1, FILE *f1, char *file2, FILE *f2, int flags)
863 int m, i0, i1, j0, j1;
869 J[m + 1] = len[1] + 1;
870 if (diff_format != D_EDIT) {
871 for (i0 = 1; i0 <= m; i0 = i1 + 1) {
872 while (i0 <= m && J[i0] == J[i0 - 1] + 1)
876 while (i1 < m && J[i1 + 1] == 0)
880 change(file1, f1, file2, f2, i0, i1, j0, j1, &flags);
883 for (i0 = m; i0 >= 1; i0 = i1 - 1) {
884 while (i0 >= 1 && J[i0] == J[i0 + 1] - 1 && J[i0] != 0)
888 while (i1 > 1 && J[i1 - 1] == 0)
892 change(file1, f1, file2, f2, i1, i0, j1, j0, &flags);
896 change(file1, f1, file2, f2, 1, 0, 1, len[1], &flags);
897 if (diff_format == D_IFDEF || diff_format == D_GFORMAT) {
900 if ((c = getc(f1)) == EOF)
902 diff_output("%c", c);
906 if (anychange != 0) {
907 if (diff_format == D_CONTEXT)
908 dump_context_vec(f1, f2, flags);
909 else if (diff_format == D_UNIFIED)
910 dump_unified_vec(f1, f2, flags);
915 range(int a, int b, const char *separator)
917 diff_output("%d", a > b ? b : a);
919 diff_output("%s%d", separator, b);
923 uni_range(int a, int b)
926 diff_output("%d,%d", a, b - a + 1);
928 diff_output("%d", b);
930 diff_output("%d,0", b);
934 preadline(int fd, size_t rlen, off_t off)
939 line = xmalloc(rlen + 1);
940 if ((nr = pread(fd, line, rlen, off)) < 0)
942 if (nr > 0 && line[nr-1] == '\n')
949 ignoreline_pattern(char *line)
953 ret = regexec(&ignore_re, line, 0, NULL, 0);
955 return (ret == 0); /* if it matched, it should be ignored. */
959 ignoreline(char *line, bool skip_blanks)
962 if (ignore_pats != NULL && skip_blanks)
963 return (ignoreline_pattern(line) || *line == '\0');
964 if (ignore_pats != NULL)
965 return (ignoreline_pattern(line));
967 return (*line == '\0');
968 /* No ignore criteria specified */
973 * Indicate that there is a difference between lines a and b of the from file
974 * to get to lines c to d of the to file. If a is greater then b then there
975 * are no lines in the from file involved and this means that there were
976 * lines appended (beginning at b). If c is greater than d then there are
977 * lines missing from the to file.
980 change(char *file1, FILE *f1, char *file2, FILE *f2, int a, int b, int c, int d,
983 static size_t max_context = 64;
989 skip_blanks = (*pflags & D_SKIPBLANKLINES);
991 if ((diff_format != D_IFDEF || diff_format == D_GFORMAT) &&
994 if (ignore_pats != NULL || skip_blanks) {
997 * All lines in the change, insert, or delete must
998 * match an ignore pattern for the change to be
1001 if (a <= b) { /* Changes and deletes. */
1002 for (i = a; i <= b; i++) {
1003 line = preadline(fileno(f1),
1004 ixold[i] - ixold[i - 1], ixold[i - 1]);
1005 if (!ignoreline(line, skip_blanks))
1009 if (a > b || c <= d) { /* Changes and inserts. */
1010 for (i = c; i <= d; i++) {
1011 line = preadline(fileno(f2),
1012 ixnew[i] - ixnew[i - 1], ixnew[i - 1]);
1013 if (!ignoreline(line, skip_blanks))
1020 if (*pflags & D_HEADER && diff_format != D_BRIEF) {
1021 diff_output("%s %s %s\n", diffargs, file1, file2);
1022 *pflags &= ~D_HEADER;
1024 if (diff_format == D_CONTEXT || diff_format == D_UNIFIED) {
1026 * Allocate change records as needed.
1028 if (context_vec_ptr == context_vec_end - 1) {
1029 ptrdiff_t offset = context_vec_ptr - context_vec_start;
1031 context_vec_start = xreallocarray(context_vec_start,
1032 max_context, sizeof(*context_vec_start));
1033 context_vec_end = context_vec_start + max_context;
1034 context_vec_ptr = context_vec_start + offset;
1036 if (anychange == 0) {
1038 * Print the context/unidiff header first time through.
1040 print_header(file1, file2);
1042 } else if (a > context_vec_ptr->b + (2 * diff_context) + 1 &&
1043 c > context_vec_ptr->d + (2 * diff_context) + 1) {
1045 * If this change is more than 'diff_context' lines from the
1046 * previous change, dump the record and reset it.
1048 if (diff_format == D_CONTEXT)
1049 dump_context_vec(f1, f2, *pflags);
1051 dump_unified_vec(f1, f2, *pflags);
1054 context_vec_ptr->a = a;
1055 context_vec_ptr->b = b;
1056 context_vec_ptr->c = c;
1057 context_vec_ptr->d = d;
1062 switch (diff_format) {
1068 diff_output("%c", a > b ? 'a' : c > d ? 'd' : 'c');
1069 if (diff_format == D_NORMAL)
1074 diff_output("%c", a > b ? 'a' : c > d ? 'd' : 'c');
1080 diff_output("a%d %d\n", b, d - c + 1);
1082 diff_output("d%d %d\n", a, b - a + 1);
1084 /* add changed lines */
1085 diff_output("a%d %d\n", b, d - c + 1);
1089 if (diff_format == D_GFORMAT) {
1091 /* print through if append (a>b), else to (nb: 0 vs 1 orig) */
1092 nc = ixold[a > b ? b : a - 1] - curpos;
1093 for (i = 0; i < nc; i++)
1094 diff_output("%c", getc(f1));
1095 for (walk = group_format; *walk != '\0'; walk++) {
1100 fetch(ixold, a, b, f1, '<', 1, *pflags);
1103 fetch(ixnew, c, d, f2, '>', 0, *pflags);
1106 diff_output("%%%c", *walk);
1111 diff_output("%c", *walk);
1114 if (diff_format == D_NORMAL || diff_format == D_IFDEF) {
1115 fetch(ixold, a, b, f1, '<', 1, *pflags);
1116 if (a <= b && c <= d && diff_format == D_NORMAL)
1117 diff_output("---\n");
1120 if (diff_format != D_GFORMAT)
1121 f = fetch(ixnew, c, d, f2, diff_format == D_NORMAL ? '>' : '\0', 0, *pflags);
1122 if (f != 0 && diff_format == D_EDIT) {
1124 * A non-zero return value for D_EDIT indicates that the
1125 * last line printed was a bare dot (".") that has been
1126 * escaped as ".." to prevent ed(1) from misinterpreting
1127 * it. We have to add a substitute command to change this
1128 * back and restart where we left off.
1131 diff_output("%ds/.//\n", a + f - 1);
1137 if ((diff_format == D_EDIT || diff_format == D_REVERSE) && c <= d)
1140 diff_output("#endif /* %s */\n", ifdefname);
1146 fetch(long *f, int a, int b, FILE *lb, int ch, int oldfile, int flags)
1148 int i, j, c, lastc, col, nc;
1152 * When doing #ifdef's, copy down to current line
1153 * if this is the first file, so that stuff makes it to output.
1155 if ((diff_format == D_IFDEF) && oldfile) {
1156 long curpos = ftell(lb);
1157 /* print through if append (a>b), else to (nb: 0 vs 1 orig) */
1158 nc = f[a > b ? b : a - 1] - curpos;
1159 for (i = 0; i < nc; i++)
1160 diff_output("%c", getc(lb));
1164 if (diff_format == D_IFDEF) {
1166 diff_output("#else /* %s%s */\n",
1167 oldfile == 1 ? "!" : "", ifdefname);
1170 diff_output("#ifndef %s\n", ifdefname);
1172 diff_output("#ifdef %s\n", ifdefname);
1174 inifdef = 1 + oldfile;
1176 for (i = a; i <= b; i++) {
1177 fseek(lb, f[i - 1], SEEK_SET);
1178 nc = f[i] - f[i - 1];
1179 if ((diff_format != D_IFDEF && diff_format != D_GFORMAT) &&
1181 diff_output("%c", ch);
1182 if (Tflag && (diff_format == D_NORMAL || diff_format == D_CONTEXT
1183 || diff_format == D_UNIFIED))
1185 else if (diff_format != D_UNIFIED)
1189 for (j = 0, lastc = '\0'; j < nc; j++, lastc = c) {
1190 if ((c = getc(lb)) == EOF) {
1191 if (diff_format == D_EDIT || diff_format == D_REVERSE ||
1192 diff_format == D_NREVERSE)
1193 warnx("No newline at end of file");
1195 diff_output("\n\\ No newline at end of "
1199 if (c == '\t' && (flags & D_EXPANDTABS)) {
1200 newcol = ((col/tabsize)+1)*tabsize;
1203 } while (++col < newcol);
1205 if (diff_format == D_EDIT && j == 1 && c == '\n'
1208 * Don't print a bare "." line
1209 * since that will confuse ed(1).
1210 * Print ".." instead and return,
1211 * giving the caller an offset
1212 * from which to restart.
1217 diff_output("%c", c);
1226 * Hash function taken from Robert Sedgewick, Algorithms in C, 3d ed., p 578.
1229 readhash(FILE *f, int flags)
1236 if ((flags & (D_FOLDBLANKS|D_IGNOREBLANKS)) == 0) {
1237 if (flags & D_IGNORECASE)
1238 for (i = 0; (t = getc(f)) != '\n'; i++) {
1239 if (flags & D_STRIPCR && t == '\r') {
1250 sum = sum * 127 + chrtran(t);
1253 for (i = 0; (t = getc(f)) != '\n'; i++) {
1254 if (flags & D_STRIPCR && t == '\r') {
1265 sum = sum * 127 + t;
1269 switch (t = getc(f)) {
1278 if (space && (flags & D_IGNOREBLANKS) == 0) {
1282 sum = sum * 127 + chrtran(t);
1296 * There is a remote possibility that we end up with a zero sum.
1297 * Zero is used as an EOF marker, so return 1 instead.
1299 return (sum == 0 ? 1 : sum);
1305 unsigned char buf[BUFSIZ];
1312 cnt = fread(buf, 1, sizeof(buf), f);
1313 return (memchr(buf, '\0', cnt) == NULL);
1316 #define begins_with(s, pre) (strncmp(s, pre, sizeof(pre)-1) == 0)
1319 match_function(const long *f, int pos, FILE *fp)
1321 unsigned char buf[FUNCTION_CONTEXT_SIZE];
1323 int last = lastline;
1324 const char *state = NULL;
1327 while (pos > last) {
1328 fseek(fp, f[pos - 1], SEEK_SET);
1329 nc = f[pos] - f[pos - 1];
1330 if (nc >= sizeof(buf))
1331 nc = sizeof(buf) - 1;
1332 nc = fread(buf, 1, nc, fp);
1335 buf[strcspn(buf, "\n")] = '\0';
1336 if (isalpha(buf[0]) || buf[0] == '_' || buf[0] == '$') {
1337 if (begins_with(buf, "private:")) {
1339 state = " (private)";
1340 } else if (begins_with(buf, "protected:")) {
1342 state = " (protected)";
1343 } else if (begins_with(buf, "public:")) {
1345 state = " (public)";
1347 strlcpy(lastbuf, buf, sizeof lastbuf);
1349 strlcat(lastbuf, state,
1351 lastmatchline = pos;
1358 return lastmatchline > 0 ? lastbuf : NULL;
1361 /* dump accumulated "context" diff changes */
1363 dump_context_vec(FILE *f1, FILE *f2, int flags)
1365 struct context_vec *cvp = context_vec_start;
1366 int lowa, upb, lowc, upd, do_output;
1370 if (context_vec_start > context_vec_ptr)
1373 b = d = 0; /* gcc */
1374 lowa = MAX(1, cvp->a - diff_context);
1375 upb = MIN(len[0], context_vec_ptr->b + diff_context);
1376 lowc = MAX(1, cvp->c - diff_context);
1377 upd = MIN(len[1], context_vec_ptr->d + diff_context);
1379 diff_output("***************");
1380 if ((flags & D_PROTOTYPE)) {
1381 f = match_function(ixold, lowa-1, f1);
1383 diff_output(" %s", f);
1385 diff_output("\n*** ");
1386 range(lowa, upb, ",");
1387 diff_output(" ****\n");
1390 * Output changes to the "old" file. The first loop suppresses
1391 * output if there were no changes to the "old" file (we'll see
1392 * the "old" lines as context in the "new" list).
1395 for (; cvp <= context_vec_ptr; cvp++)
1396 if (cvp->a <= cvp->b) {
1397 cvp = context_vec_start;
1402 while (cvp <= context_vec_ptr) {
1408 if (a <= b && c <= d)
1411 ch = (a <= b) ? 'd' : 'a';
1414 fetch(ixold, lowa, b, f1, ' ', 0, flags);
1416 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags);
1417 fetch(ixold, a, b, f1,
1418 ch == 'c' ? '!' : '-', 0, flags);
1423 fetch(ixold, b + 1, upb, f1, ' ', 0, flags);
1425 /* output changes to the "new" file */
1426 diff_output("--- ");
1427 range(lowc, upd, ",");
1428 diff_output(" ----\n");
1431 for (cvp = context_vec_start; cvp <= context_vec_ptr; cvp++)
1432 if (cvp->c <= cvp->d) {
1433 cvp = context_vec_start;
1438 while (cvp <= context_vec_ptr) {
1444 if (a <= b && c <= d)
1447 ch = (a <= b) ? 'd' : 'a';
1450 fetch(ixnew, lowc, d, f2, ' ', 0, flags);
1452 fetch(ixnew, lowc, c - 1, f2, ' ', 0, flags);
1453 fetch(ixnew, c, d, f2,
1454 ch == 'c' ? '!' : '+', 0, flags);
1459 fetch(ixnew, d + 1, upd, f2, ' ', 0, flags);
1461 context_vec_ptr = context_vec_start - 1;
1464 /* dump accumulated "unified" diff changes */
1466 dump_unified_vec(FILE *f1, FILE *f2, int flags)
1468 struct context_vec *cvp = context_vec_start;
1469 int lowa, upb, lowc, upd;
1473 if (context_vec_start > context_vec_ptr)
1476 b = d = 0; /* gcc */
1477 lowa = MAX(1, cvp->a - diff_context);
1478 upb = MIN(len[0], context_vec_ptr->b + diff_context);
1479 lowc = MAX(1, cvp->c - diff_context);
1480 upd = MIN(len[1], context_vec_ptr->d + diff_context);
1482 diff_output("@@ -");
1483 uni_range(lowa, upb);
1485 uni_range(lowc, upd);
1487 if ((flags & D_PROTOTYPE)) {
1488 f = match_function(ixold, lowa-1, f1);
1490 diff_output(" %s", f);
1495 * Output changes in "unified" diff format--the old and new lines
1496 * are printed together.
1498 for (; cvp <= context_vec_ptr; cvp++) {
1505 * c: both new and old changes
1506 * d: only changes in the old file
1507 * a: only changes in the new file
1509 if (a <= b && c <= d)
1512 ch = (a <= b) ? 'd' : 'a';
1516 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags);
1517 fetch(ixold, a, b, f1, '-', 0, flags);
1518 fetch(ixnew, c, d, f2, '+', 0, flags);
1521 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags);
1522 fetch(ixold, a, b, f1, '-', 0, flags);
1525 fetch(ixnew, lowc, c - 1, f2, ' ', 0, flags);
1526 fetch(ixnew, c, d, f2, '+', 0, flags);
1532 fetch(ixnew, d + 1, upd, f2, ' ', 0, flags);
1534 context_vec_ptr = context_vec_start - 1;
1538 print_header(const char *file1, const char *file2)
1540 const char *time_format;
1545 struct tm tm1, tm2, *tm_ptr1, *tm_ptr2;
1546 int nsec1 = stb1.st_mtim.tv_nsec;
1547 int nsec2 = stb2.st_mtim.tv_nsec;
1549 time_format = "%Y-%m-%d %H:%M:%S";
1553 tm_ptr1 = localtime_r(&stb1.st_mtime, &tm1);
1554 tm_ptr2 = localtime_r(&stb2.st_mtime, &tm2);
1555 strftime(buf1, 256, time_format, tm_ptr1);
1556 strftime(buf2, 256, time_format, tm_ptr2);
1558 strftime(end1, 10, "%z", tm_ptr1);
1559 strftime(end2, 10, "%z", tm_ptr2);
1560 sprintf(buf1, "%s.%.9d %s", buf1, nsec1, end1);
1561 sprintf(buf2, "%s.%.9d %s", buf2, nsec2, end2);
1563 if (label[0] != NULL)
1564 diff_output("%s %s\n", diff_format == D_CONTEXT ? "***" : "---",
1567 diff_output("%s %s\t%s\n", diff_format == D_CONTEXT ? "***" : "---",
1569 if (label[1] != NULL)
1570 diff_output("%s %s\n", diff_format == D_CONTEXT ? "---" : "+++",
1573 diff_output("%s %s\t%s\n", diff_format == D_CONTEXT ? "---" : "+++",