3 * finddup is a simple utility to find duplicated files, files common
4 * to several directories, or files present in one directory and not
7 * Copyright (c) 2010 Francois Fleuret
8 * Written by Francois Fleuret <francois@fleuret.org>
10 * This file is part of finddup.
12 * finddup is free software: you can redistribute it and/or modify it
13 * under the terms of the GNU General Public License version 3 as
14 * published by the Free Software Foundation.
16 * finddup is distributed in the hope that it will be useful, but WITHOUT
17 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
18 * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public
19 * License for more details.
21 * You should have received a copy of the GNU General Public License
22 * along with finddup. If not, see <http://www.gnu.org/licenses/>.
26 #define VERSION_NUMBER "0.9"
30 #include <sys/types.h>
32 #include <sys/param.h>
39 #include <sys/ioctl.h>
43 #include <openssl/md5.h>
45 /* 1M really helps compared to 64k */
46 #define READ_BUFFER_SIZE (1024 * 1024)
48 typedef int64_t size_sum_t;
50 /* Yeah, global variables! */
52 int ignore_dotfiles = 0; /* 1 means ignore files and directories
53 starting with a dot */
55 int ignore_empty_files = 0; /* 1 means ignore empty files */
57 int show_realpaths = 0; /* 1 means ignore files and directories
58 starting with a dot */
60 int show_progress = 0; /* 1 means show a progress bar when we are in a
63 int show_hits = 1; /* 1 means to show the files from dir2
64 corresponding to the ones from dir1 */
66 int show_groups = 1; /* 1 means to show the group IDs when printing
69 int ignore_same_inode = 0; /* 1 means that comparison between two file
70 with same inode will always be false */
72 int tty_width = -1; /* Positive value means what width to use to show
75 int use_md5 = 0; /* 1 means we keep an MD5 signature for each file */
77 /********************************************************************/
79 /* malloc with error checking. */
81 void *safe_malloc(size_t n) {
84 printf("Can not allocate memory: %s\n", strerror(errno));
90 /********************************************************************/
92 int ignore_entry(const char *name) {
94 strcmp(name, ".") == 0 ||
95 strcmp(name, "..") == 0 ||
96 (ignore_dotfiles && name[0] == '.' && name[1] != '/');
99 /**********************************************************************/
102 struct file_node *next;
106 int group_id; /* one per identical file content */
107 int dir_id; /* 1 for DIR1, and 2 for DIR2 */
109 unsigned char md5[MD5_DIGEST_LENGTH];
112 void file_list_delete(struct file_node *head) {
113 struct file_node *next;
122 int file_list_length(struct file_node *head) {
131 /**********************************************************************/
133 int same_content(struct file_node *f1, struct file_node *f2,
134 char *buffer1, char *buffer2) {
135 int fd1, fd2, s1, s2;
139 if(f1->md5_computed && f2->md5_computed &&
140 !memcmp(f1->md5, f2->md5, MD5_DIGEST_LENGTH)) {
143 if(!f1->md5_computed) {
146 if(!f2->md5_computed) {
152 fd1 = open(f1->name, O_RDONLY);
153 fd2 = open(f2->name, O_RDONLY);
155 if(fd1 >= 0 && fd2 >= 0) {
157 s1 = read(fd1, buffer1, READ_BUFFER_SIZE);
158 s2 = read(fd2, buffer2, READ_BUFFER_SIZE);
160 if(s1 < 0 || s2 < 0) {
171 if(!f1->md5_computed) {
172 MD5_Final(f1->md5, &c1);
173 f1->md5_computed = 1;
175 if(!f2->md5_computed) {
176 MD5_Final(f2->md5, &c2);
177 f2->md5_computed = 1;
183 if(!f1->md5_computed) {
184 MD5_Update(&c1, buffer1, s1);
186 if(!f2->md5_computed) {
187 MD5_Update(&c2, buffer2, s2);
190 if(memcmp(buffer1, buffer2, s1)) {
198 "Different read size without error on files of same size.\n");
206 "Can not open \"%s\" error: %s\n",
213 "Can not open \"%s\" error: %s\n",
222 int same_files(struct file_node *f1, struct file_node *f2,
223 char *buffer1, char *buffer2) {
224 if(ignore_same_inode && f1->inode == f2->inode) {
227 return f1->size == f2->size && same_content(f1, f2, buffer1, buffer2);
230 /**********************************************************************/
232 struct file_node *scan_directory(struct file_node *tail, const char *name) {
234 struct dirent *dir_e;
236 struct file_node *tmp;
237 char subname[PATH_MAX + 1];
239 if(lstat(name, &sb) != 0) {
240 fprintf(stderr, "Can not stat \"%s\": %s\n", name, strerror(errno));
244 if(S_ISLNK(sb.st_mode)) {
251 while((dir_e = readdir(dir))) {
252 if(!ignore_entry(dir_e->d_name)) {
253 snprintf(subname, PATH_MAX, "%s/%s", name, dir_e->d_name);
254 tail = scan_directory(tail, subname);
259 if(S_ISREG(sb.st_mode)) {
260 if(!ignore_entry(name)) {
261 if(!ignore_empty_files || sb.st_size > 0) {
262 tmp = safe_malloc(sizeof(struct file_node));
264 tmp->name = strdup(name);
265 tmp->size = sb.st_size;
266 tmp->inode = sb.st_ino;
269 tmp->md5_computed = 0;
279 void print_file(struct file_node *node) {
280 char tmp[PATH_MAX + 1];
282 if(realpath(node->name, tmp)) {
284 printf("%d %s\n", node->group_id, tmp);
289 printf("Can not get the realpath of \"%s\": %s\n",
296 printf("%d %s\n", node->group_id, node->name);
298 printf("%s\n", node->name);
303 int compare_nodes(const void *x1, const void *x2) {
304 const struct file_node **f1, **f2;
306 f1 = (const struct file_node **) x1;
307 f2 = (const struct file_node **) x2;
309 if((*f1)->group_id < (*f2)->group_id) {
311 } else if((*f1)->group_id > (*f2)->group_id) {
314 if((*f1)->dir_id < (*f2)->dir_id) {
316 } else if((*f1)->dir_id > (*f2)->dir_id) {
325 void print_result(struct file_node *list1, struct file_node *list2) {
326 struct file_node *node1, *node2;
327 struct file_node **nodes;
331 for(node1 = list1; node1; node1 = node1->next) {
332 if(node1->group_id >= 0) { nb++; }
336 for(node2 = list2; node2; node2 = node2->next) {
337 if(node2->group_id >= 0) { nb++; }
341 nodes = safe_malloc(nb * sizeof(struct file_node *));
344 for(node1 = list1; node1; node1 = node1->next) {
345 if(node1->group_id >= 0) {
351 for(node2 = list2; node2; node2 = node2->next) {
352 if(node2->group_id >= 0) {
358 qsort(nodes, nb, sizeof(struct file_node *), compare_nodes);
360 for(n = 0; n < nb; n++) {
361 if(!show_groups && n > 0 && nodes[n]->group_id != nodes[n-1]->group_id) {
364 print_file(nodes[n]);
370 void print_progress(int max, int n, int *pp) {
373 if(show_progress && tty_width > 0) {
374 width = tty_width - 7;
375 p = (width * n) / (max - 1);
377 for(k = 0; k < p; k++) {
378 fprintf(stderr, "+");
380 for(; k < width; k++) {
381 fprintf(stderr, "-");
384 p = (100 * n) / (max - 1);
385 fprintf(stderr, " [%3d%%]\r", p);
390 void start(const char *dirname1, const char *dirname2) {
391 struct file_node *list1, *list2;
392 struct file_node *node1, *node2;
394 int nb_groups, nb_nodes;
395 int list1_length, previous_progress;
398 char *buffer1 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
399 char *buffer2 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
404 if(isatty(STDOUT_FILENO) &&
405 !ioctl (STDOUT_FILENO, TIOCGWINSZ, (char *) &win)) {
406 tty_width = win.ws_col;
408 fprintf(stderr, "Scanning %s ... ", dirname1);
411 list1 = scan_directory(0, dirname1);
414 if(strncmp(dirname2, "not:", 4) == 0) {
416 /* groups are not computed in the not: mode */
419 } else if(strncmp(dirname2, "and:", 4) == 0) {
423 fprintf(stderr, "%s ... ", dirname2);
425 list2 = scan_directory(0, dirname2);
431 fprintf(stderr, "done.\n");
435 previous_progress = -1;
437 list1_length = file_list_length(list1);
440 for(node1 = list1; node1; node1 = node1->next) {
441 print_progress(list1_length, nb_nodes, &previous_progress);
446 for(node2 = list2; !found && node2; node2 = node2->next) {
447 if(same_files(node1, node2, buffer1, buffer2)) {
454 printf("%s\n", realpath(node1->name, 0));
456 printf("%s\n", node1->name);
462 for(node1 = list1; node1; node1 = node1->next) {
463 print_progress(list1_length, nb_nodes, &previous_progress);
466 for(node2 = list2; node2; node2 = node2->next) {
467 if(node1->group_id < 0 || node2->group_id < 0) {
468 if(same_files(node1, node2, buffer1, buffer2)) {
469 if(node1->group_id < 0) {
470 if(node2->group_id >= 0) {
471 node1->group_id = node2->group_id;
473 node1->group_id = nb_groups;
478 if(node2->group_id < 0) {
479 node2->group_id = node1->group_id;
489 fprintf(stderr, "\n");
493 print_result(list1, list2);
494 file_list_delete(list1);
495 file_list_delete(list2);
497 print_result(list1, 0);
498 file_list_delete(list1);
505 void print_help(FILE *out) {
506 fprintf(out, "Usage: finddup [OPTION]... DIR1 [[and:|not:]DIR2]\n");
507 fprintf(out, "Version %s (%s)\n", VERSION_NUMBER, UNAME);
508 fprintf(out, "Without DIR2, lists duplicated files found in DIR1. With DIR2, lists files common to both directories. With the not: prefix, lists files found in DIR1 which do not exist in DIR2. The and: prefix is the default and should be used only if you have a directory starting with 'not:'\n");
510 /* 01234567890123456789012345678901234567890123456789012345678901234567890123456789*/
511 fprintf(out, " -h, --help show this help\n");
512 fprintf(out, " -d, --ignore-dots ignore dot files and directories\n");
513 fprintf(out, " -0, --ignore-empty ignore empty files\n");
514 fprintf(out, " -c, --hide-matchings do not show which files in DIR2 corresponds to\n");
515 fprintf(out, " those in DIR1\n");
516 fprintf(out, " -g, --no-group-ids do not show the file groups\n");
517 fprintf(out, " -p, --show-progress show progress\n");
518 fprintf(out, " -r, --real-paths show the real file paths\n");
519 fprintf(out, " -i, --same-inodes-are-different\n");
520 fprintf(out, " consider files with same inode as different\n");
521 fprintf(out, " -m, --md5 use MD5 hashing\n");
523 fprintf(out, "Report bugs and comments to <francois@fleuret.org>.\n");
526 /**********************************************************************/
528 static struct option long_options[] = {
529 { "help", no_argument, 0, 'h' },
530 { "same-inodes-are-different", no_argument, 0, 'i' },
531 { "real-paths", no_argument, 0, 'r' },
532 { "hide-matchings", no_argument, 0, 'c' },
533 { "no-group-ids", no_argument, 0, 'g' },
534 { "ignore-dots", no_argument, 0, 'd' },
535 { "ignore-empty", no_argument, 0, '0' },
536 { "show-progress", no_argument, 0, 'p' },
537 { "md5", no_argument, 0, 'm' },
541 int main(int argc, char **argv) {
544 setlocale (LC_ALL, "");
546 while ((c = getopt_long(argc, argv, "hircgd0pm",
547 long_options, NULL)) != -1) {
561 ignore_empty_files = 1;
569 ignore_same_inode = 1;
593 if(optind + 2 == argc) {
594 start(argv[optind], argv[optind + 1]);
595 } else if(optind + 1 == argc) {
596 ignore_same_inode = 1;
597 start(argv[optind], 0);