1*c9733229SMatthew Dillon /* $OpenBSD: diffdir.c,v 1.47 2019/01/25 00:19:26 millert Exp $ */
2*c9733229SMatthew Dillon
3*c9733229SMatthew Dillon /*
4*c9733229SMatthew Dillon * Copyright (c) 2003, 2010 Todd C. Miller <millert@openbsd.org>
5*c9733229SMatthew Dillon *
6*c9733229SMatthew Dillon * Permission to use, copy, modify, and distribute this software for any
7*c9733229SMatthew Dillon * purpose with or without fee is hereby granted, provided that the above
8*c9733229SMatthew Dillon * copyright notice and this permission notice appear in all copies.
9*c9733229SMatthew Dillon *
10*c9733229SMatthew Dillon * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11*c9733229SMatthew Dillon * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12*c9733229SMatthew Dillon * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13*c9733229SMatthew Dillon * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14*c9733229SMatthew Dillon * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15*c9733229SMatthew Dillon * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16*c9733229SMatthew Dillon * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17*c9733229SMatthew Dillon *
18*c9733229SMatthew Dillon * Sponsored in part by the Defense Advanced Research Projects
19*c9733229SMatthew Dillon * Agency (DARPA) and Air Force Research Laboratory, Air Force
20*c9733229SMatthew Dillon * Materiel Command, USAF, under agreement number F39502-99-1-0512.
21*c9733229SMatthew Dillon */
22*c9733229SMatthew Dillon
23*c9733229SMatthew Dillon #include <sys/stat.h>
24*c9733229SMatthew Dillon
25*c9733229SMatthew Dillon #include <dirent.h>
26*c9733229SMatthew Dillon #include <err.h>
27*c9733229SMatthew Dillon #include <errno.h>
28*c9733229SMatthew Dillon #include <fcntl.h>
29*c9733229SMatthew Dillon #include <fnmatch.h>
30*c9733229SMatthew Dillon #include <paths.h>
31*c9733229SMatthew Dillon #include <stdio.h>
32*c9733229SMatthew Dillon #include <stdlib.h>
33*c9733229SMatthew Dillon #include <string.h>
34*c9733229SMatthew Dillon #include <unistd.h>
35*c9733229SMatthew Dillon #include <limits.h>
36*c9733229SMatthew Dillon
37*c9733229SMatthew Dillon #include "diff.h"
38*c9733229SMatthew Dillon #include "xmalloc.h"
39*c9733229SMatthew Dillon
40*c9733229SMatthew Dillon static int selectfile(const struct dirent *);
41*c9733229SMatthew Dillon static void diffit(struct dirent *, char *, size_t, char *, size_t, int);
42*c9733229SMatthew Dillon
43*c9733229SMatthew Dillon #define d_status d_type /* we need to store status for -l */
44*c9733229SMatthew Dillon
45*c9733229SMatthew Dillon /*
46*c9733229SMatthew Dillon * Diff directory traversal. Will be called recursively if -r was specified.
47*c9733229SMatthew Dillon */
48*c9733229SMatthew Dillon void
diffdir(char * p1,char * p2,int flags)49*c9733229SMatthew Dillon diffdir(char *p1, char *p2, int flags)
50*c9733229SMatthew Dillon {
51*c9733229SMatthew Dillon struct dirent *dent1, **dp1, **edp1 = NULL, **dirp1 = NULL;
52*c9733229SMatthew Dillon struct dirent *dent2, **dp2, **edp2 = NULL, **dirp2 = NULL;
53*c9733229SMatthew Dillon size_t dirlen1, dirlen2;
54*c9733229SMatthew Dillon char path1[PATH_MAX], path2[PATH_MAX];
55*c9733229SMatthew Dillon int pos;
56*c9733229SMatthew Dillon
57*c9733229SMatthew Dillon dirlen1 = strlcpy(path1, *p1 ? p1 : ".", sizeof(path1));
58*c9733229SMatthew Dillon if (dirlen1 >= sizeof(path1) - 1) {
59*c9733229SMatthew Dillon warnc(ENAMETOOLONG, "%s", p1);
60*c9733229SMatthew Dillon status |= 2;
61*c9733229SMatthew Dillon return;
62*c9733229SMatthew Dillon }
63*c9733229SMatthew Dillon if (path1[dirlen1 - 1] != '/') {
64*c9733229SMatthew Dillon path1[dirlen1++] = '/';
65*c9733229SMatthew Dillon path1[dirlen1] = '\0';
66*c9733229SMatthew Dillon }
67*c9733229SMatthew Dillon dirlen2 = strlcpy(path2, *p2 ? p2 : ".", sizeof(path2));
68*c9733229SMatthew Dillon if (dirlen2 >= sizeof(path2) - 1) {
69*c9733229SMatthew Dillon warnc(ENAMETOOLONG, "%s", p2);
70*c9733229SMatthew Dillon status |= 2;
71*c9733229SMatthew Dillon return;
72*c9733229SMatthew Dillon }
73*c9733229SMatthew Dillon if (path2[dirlen2 - 1] != '/') {
74*c9733229SMatthew Dillon path2[dirlen2++] = '/';
75*c9733229SMatthew Dillon path2[dirlen2] = '\0';
76*c9733229SMatthew Dillon }
77*c9733229SMatthew Dillon
78*c9733229SMatthew Dillon /*
79*c9733229SMatthew Dillon * Get a list of entries in each directory, skipping "excluded" files
80*c9733229SMatthew Dillon * and sorting alphabetically.
81*c9733229SMatthew Dillon */
82*c9733229SMatthew Dillon pos = scandir(path1, &dirp1, selectfile, alphasort);
83*c9733229SMatthew Dillon if (pos == -1) {
84*c9733229SMatthew Dillon if (errno == ENOENT && (Nflag || Pflag)) {
85*c9733229SMatthew Dillon pos = 0;
86*c9733229SMatthew Dillon } else {
87*c9733229SMatthew Dillon warn("%s", path1);
88*c9733229SMatthew Dillon goto closem;
89*c9733229SMatthew Dillon }
90*c9733229SMatthew Dillon }
91*c9733229SMatthew Dillon dp1 = dirp1;
92*c9733229SMatthew Dillon edp1 = dirp1 + pos;
93*c9733229SMatthew Dillon
94*c9733229SMatthew Dillon pos = scandir(path2, &dirp2, selectfile, alphasort);
95*c9733229SMatthew Dillon if (pos == -1) {
96*c9733229SMatthew Dillon if (errno == ENOENT && Nflag) {
97*c9733229SMatthew Dillon pos = 0;
98*c9733229SMatthew Dillon } else {
99*c9733229SMatthew Dillon warn("%s", path2);
100*c9733229SMatthew Dillon goto closem;
101*c9733229SMatthew Dillon }
102*c9733229SMatthew Dillon }
103*c9733229SMatthew Dillon dp2 = dirp2;
104*c9733229SMatthew Dillon edp2 = dirp2 + pos;
105*c9733229SMatthew Dillon
106*c9733229SMatthew Dillon /*
107*c9733229SMatthew Dillon * If we were given a starting point, find it.
108*c9733229SMatthew Dillon */
109*c9733229SMatthew Dillon if (start != NULL) {
110*c9733229SMatthew Dillon while (dp1 != edp1 && strcmp((*dp1)->d_name, start) < 0)
111*c9733229SMatthew Dillon dp1++;
112*c9733229SMatthew Dillon while (dp2 != edp2 && strcmp((*dp2)->d_name, start) < 0)
113*c9733229SMatthew Dillon dp2++;
114*c9733229SMatthew Dillon }
115*c9733229SMatthew Dillon
116*c9733229SMatthew Dillon /*
117*c9733229SMatthew Dillon * Iterate through the two directory lists, diffing as we go.
118*c9733229SMatthew Dillon */
119*c9733229SMatthew Dillon while (dp1 != edp1 || dp2 != edp2) {
120*c9733229SMatthew Dillon dent1 = dp1 != edp1 ? *dp1 : NULL;
121*c9733229SMatthew Dillon dent2 = dp2 != edp2 ? *dp2 : NULL;
122*c9733229SMatthew Dillon
123*c9733229SMatthew Dillon pos = dent1 == NULL ? 1 : dent2 == NULL ? -1 :
124*c9733229SMatthew Dillon strcmp(dent1->d_name, dent2->d_name);
125*c9733229SMatthew Dillon if (pos == 0) {
126*c9733229SMatthew Dillon /* file exists in both dirs, diff it */
127*c9733229SMatthew Dillon diffit(dent1, path1, dirlen1, path2, dirlen2, flags);
128*c9733229SMatthew Dillon dp1++;
129*c9733229SMatthew Dillon dp2++;
130*c9733229SMatthew Dillon } else if (pos < 0) {
131*c9733229SMatthew Dillon /* file only in first dir, only diff if -N */
132*c9733229SMatthew Dillon if (Nflag) {
133*c9733229SMatthew Dillon diffit(dent1, path1, dirlen1, path2, dirlen2,
134*c9733229SMatthew Dillon flags);
135*c9733229SMatthew Dillon } else {
136*c9733229SMatthew Dillon print_only(path1, dirlen1, dent1->d_name);
137*c9733229SMatthew Dillon status |= 1;
138*c9733229SMatthew Dillon }
139*c9733229SMatthew Dillon dp1++;
140*c9733229SMatthew Dillon } else {
141*c9733229SMatthew Dillon /* file only in second dir, only diff if -N or -P */
142*c9733229SMatthew Dillon if (Nflag || Pflag) {
143*c9733229SMatthew Dillon diffit(dent2, path1, dirlen1, path2, dirlen2,
144*c9733229SMatthew Dillon flags);
145*c9733229SMatthew Dillon } else {
146*c9733229SMatthew Dillon print_only(path2, dirlen2, dent2->d_name);
147*c9733229SMatthew Dillon status |= 1;
148*c9733229SMatthew Dillon }
149*c9733229SMatthew Dillon dp2++;
150*c9733229SMatthew Dillon }
151*c9733229SMatthew Dillon }
152*c9733229SMatthew Dillon
153*c9733229SMatthew Dillon closem:
154*c9733229SMatthew Dillon if (dirp1 != NULL) {
155*c9733229SMatthew Dillon for (dp1 = dirp1; dp1 < edp1; dp1++)
156*c9733229SMatthew Dillon free(*dp1);
157*c9733229SMatthew Dillon free(dirp1);
158*c9733229SMatthew Dillon }
159*c9733229SMatthew Dillon if (dirp2 != NULL) {
160*c9733229SMatthew Dillon for (dp2 = dirp2; dp2 < edp2; dp2++)
161*c9733229SMatthew Dillon free(*dp2);
162*c9733229SMatthew Dillon free(dirp2);
163*c9733229SMatthew Dillon }
164*c9733229SMatthew Dillon }
165*c9733229SMatthew Dillon
166*c9733229SMatthew Dillon /*
167*c9733229SMatthew Dillon * Do the actual diff by calling either diffreg() or diffdir().
168*c9733229SMatthew Dillon */
169*c9733229SMatthew Dillon static void
diffit(struct dirent * dp,char * path1,size_t plen1,char * path2,size_t plen2,int flags)170*c9733229SMatthew Dillon diffit(struct dirent *dp, char *path1, size_t plen1, char *path2, size_t plen2,
171*c9733229SMatthew Dillon int flags)
172*c9733229SMatthew Dillon {
173*c9733229SMatthew Dillon flags |= D_HEADER;
174*c9733229SMatthew Dillon strlcpy(path1 + plen1, dp->d_name, PATH_MAX - plen1);
175*c9733229SMatthew Dillon if (stat(path1, &stb1) != 0) {
176*c9733229SMatthew Dillon if (!(Nflag || Pflag) || errno != ENOENT) {
177*c9733229SMatthew Dillon warn("%s", path1);
178*c9733229SMatthew Dillon return;
179*c9733229SMatthew Dillon }
180*c9733229SMatthew Dillon flags |= D_EMPTY1;
181*c9733229SMatthew Dillon memset(&stb1, 0, sizeof(stb1));
182*c9733229SMatthew Dillon }
183*c9733229SMatthew Dillon
184*c9733229SMatthew Dillon strlcpy(path2 + plen2, dp->d_name, PATH_MAX - plen2);
185*c9733229SMatthew Dillon if (stat(path2, &stb2) != 0) {
186*c9733229SMatthew Dillon if (!Nflag || errno != ENOENT) {
187*c9733229SMatthew Dillon warn("%s", path2);
188*c9733229SMatthew Dillon return;
189*c9733229SMatthew Dillon }
190*c9733229SMatthew Dillon flags |= D_EMPTY2;
191*c9733229SMatthew Dillon memset(&stb2, 0, sizeof(stb2));
192*c9733229SMatthew Dillon stb2.st_mode = stb1.st_mode;
193*c9733229SMatthew Dillon }
194*c9733229SMatthew Dillon if (stb1.st_mode == 0)
195*c9733229SMatthew Dillon stb1.st_mode = stb2.st_mode;
196*c9733229SMatthew Dillon
197*c9733229SMatthew Dillon if (S_ISDIR(stb1.st_mode) && S_ISDIR(stb2.st_mode)) {
198*c9733229SMatthew Dillon if (rflag)
199*c9733229SMatthew Dillon diffdir(path1, path2, flags);
200*c9733229SMatthew Dillon else
201*c9733229SMatthew Dillon printf("Common subdirectories: %s and %s\n",
202*c9733229SMatthew Dillon path1, path2);
203*c9733229SMatthew Dillon return;
204*c9733229SMatthew Dillon }
205*c9733229SMatthew Dillon if (!S_ISREG(stb1.st_mode) && !S_ISDIR(stb1.st_mode))
206*c9733229SMatthew Dillon dp->d_status = D_SKIPPED1;
207*c9733229SMatthew Dillon else if (!S_ISREG(stb2.st_mode) && !S_ISDIR(stb2.st_mode))
208*c9733229SMatthew Dillon dp->d_status = D_SKIPPED2;
209*c9733229SMatthew Dillon else
210*c9733229SMatthew Dillon dp->d_status = diffreg(path1, path2, flags);
211*c9733229SMatthew Dillon print_status(dp->d_status, path1, path2, "");
212*c9733229SMatthew Dillon }
213*c9733229SMatthew Dillon
214*c9733229SMatthew Dillon /*
215*c9733229SMatthew Dillon * Returns 1 if the directory entry should be included in the
216*c9733229SMatthew Dillon * diff, else 0. Checks the excludes list.
217*c9733229SMatthew Dillon */
218*c9733229SMatthew Dillon static int
selectfile(const struct dirent * dp)219*c9733229SMatthew Dillon selectfile(const struct dirent *dp)
220*c9733229SMatthew Dillon {
221*c9733229SMatthew Dillon struct excludes *excl;
222*c9733229SMatthew Dillon
223*c9733229SMatthew Dillon if (dp->d_fileno == 0)
224*c9733229SMatthew Dillon return (0);
225*c9733229SMatthew Dillon
226*c9733229SMatthew Dillon /* always skip "." and ".." */
227*c9733229SMatthew Dillon if (dp->d_name[0] == '.' && (dp->d_name[1] == '\0' ||
228*c9733229SMatthew Dillon (dp->d_name[1] == '.' && dp->d_name[2] == '\0')))
229*c9733229SMatthew Dillon return (0);
230*c9733229SMatthew Dillon
231*c9733229SMatthew Dillon /* check excludes list */
232*c9733229SMatthew Dillon for (excl = excludes_list; excl != NULL; excl = excl->next)
233*c9733229SMatthew Dillon if (fnmatch(excl->pattern, dp->d_name, FNM_PATHNAME) == 0)
234*c9733229SMatthew Dillon return (0);
235*c9733229SMatthew Dillon
236*c9733229SMatthew Dillon return (1);
237*c9733229SMatthew Dillon }
238