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>
44 #include <openssl/md5.h>
47 /* 1M really helps compared to 64k */
48 #define READ_BUFFER_SIZE (1024 * 1024)
50 #define PROGRESS_BUFFER_SIZE 1024
52 typedef int64_t size_sum_t;
54 /* Yeah, global variables! */
56 int ignore_dotfiles = 0; /* 1 means ignore files and directories
57 starting with a dot */
59 int ignore_empty_files = 0; /* 1 means ignore empty files */
61 int show_realpaths = 0; /* 1 means ignore files and directories
62 starting with a dot */
64 int show_progress = 0; /* 1 means show a progress bar when we are in a
67 int show_hits = 1; /* 1 means to show the files from dir2
68 corresponding to the ones from dir1 */
70 int show_groups = 1; /* 1 means to show the group IDs when printing
73 int same_inodes_are_different = 0; /* 1 means that comparison between
74 two files with same inode will
78 int use_md5 = 0; /* 1 means we keep an MD5 signature for each file */
81 /********************************************************************/
83 /* malloc with error checking. */
85 void *safe_malloc(size_t n) {
88 printf("Can not allocate memory: %s\n", strerror(errno));
94 /********************************************************************/
96 int ignore_entry(const char *name) {
98 strcmp(name, ".") == 0 ||
99 strcmp(name, "..") == 0 ||
100 (ignore_dotfiles && name[0] == '.' && name[1] != '/');
103 /**********************************************************************/
106 struct file_node *next;
110 int group_id; /* one per identical file content */
111 int dir_id; /* 1 for DIR1, and 2 for DIR2 */
114 unsigned char md5[MD5_DIGEST_LENGTH];
118 void file_list_delete(struct file_node *head) {
119 struct file_node *next;
128 int file_list_length(struct file_node *head) {
137 /**********************************************************************/
139 int same_content(struct file_node *f1, struct file_node *f2,
140 char *buffer1, char *buffer2) {
141 int fd1, fd2, s1, s2;
147 if(f1->md5_computed && f2->md5_computed) {
148 if(!memcmp(f1->md5, f2->md5, MD5_DIGEST_LENGTH)) {
152 if(!f1->md5_computed) {
155 if(!f2->md5_computed) {
162 fd1 = open(f1->name, O_RDONLY);
163 fd2 = open(f2->name, O_RDONLY);
165 if(fd1 >= 0 && fd2 >= 0) {
167 s1 = read(fd1, buffer1, READ_BUFFER_SIZE);
168 s2 = read(fd2, buffer2, READ_BUFFER_SIZE);
170 if(s1 < 0 || s2 < 0) {
182 if(!f1->md5_computed) {
183 MD5_Final(f1->md5, &c1);
184 f1->md5_computed = 1;
186 if(!f2->md5_computed) {
187 MD5_Final(f2->md5, &c2);
188 f2->md5_computed = 1;
194 if(memcmp(buffer1, buffer2, s1)) {
201 if(!f1->md5_computed) {
202 MD5_Update(&c1, buffer1, s1);
204 if(!f2->md5_computed) {
205 MD5_Update(&c2, buffer2, s2);
212 "Different read size without error on files of same size.\n");
220 "Can not open \"%s\" error: %s\n",
227 "Can not open \"%s\" error: %s\n",
236 int same_files(struct file_node *f1, struct file_node *f2,
237 char *buffer1, char *buffer2) {
238 if(same_inodes_are_different && f1->inode == f2->inode) {
242 return f1->size == f2->size && same_content(f1, f2, buffer1, buffer2);
245 /**********************************************************************/
247 struct file_node *scan_directory(struct file_node *tail, const char *name) {
249 struct dirent *dir_e;
251 struct file_node *tmp;
252 char subname[PATH_MAX + 1];
254 if(lstat(name, &sb) != 0) {
255 fprintf(stderr, "Can not stat \"%s\": %s\n", name, strerror(errno));
259 if(S_ISLNK(sb.st_mode)) {
266 while((dir_e = readdir(dir))) {
267 if(!ignore_entry(dir_e->d_name)) {
268 snprintf(subname, PATH_MAX, "%s/%s", name, dir_e->d_name);
269 tail = scan_directory(tail, subname);
274 if(S_ISREG(sb.st_mode)) {
275 if(!ignore_entry(name)) {
276 if(!ignore_empty_files || sb.st_size > 0) {
277 tmp = safe_malloc(sizeof(struct file_node));
279 tmp->name = strdup(name);
280 tmp->size = sb.st_size;
281 tmp->inode = sb.st_ino;
285 tmp->md5_computed = 0;
296 void print_file(struct file_node *node) {
297 char tmp[PATH_MAX + 1];
299 if(realpath(node->name, tmp)) {
301 printf("%d %s\n", node->group_id, tmp);
306 printf("Can not get the realpath of \"%s\": %s\n",
313 printf("%d %s\n", node->group_id, node->name);
315 printf("%s\n", node->name);
320 int compare_nodes(const void *x1, const void *x2) {
321 const struct file_node **f1, **f2;
323 f1 = (const struct file_node **) x1;
324 f2 = (const struct file_node **) x2;
326 if((*f1)->group_id < (*f2)->group_id) {
328 } else if((*f1)->group_id > (*f2)->group_id) {
331 if((*f1)->dir_id < (*f2)->dir_id) {
333 } else if((*f1)->dir_id > (*f2)->dir_id) {
342 void print_result(struct file_node *list1, struct file_node *list2) {
343 struct file_node *node1, *node2;
344 struct file_node **nodes;
348 for(node1 = list1; node1; node1 = node1->next) {
349 if(node1->group_id >= 0) { nb++; }
353 for(node2 = list2; node2; node2 = node2->next) {
354 if(node2->group_id >= 0) { nb++; }
358 nodes = safe_malloc(nb * sizeof(struct file_node *));
361 for(node1 = list1; node1; node1 = node1->next) {
362 if(node1->group_id >= 0) {
368 for(node2 = list2; node2; node2 = node2->next) {
369 if(node2->group_id >= 0) {
375 qsort(nodes, nb, sizeof(struct file_node *), compare_nodes);
377 for(n = 0; n < nb; n++) {
378 if(!show_groups && n > 0 && nodes[n]->group_id != nodes[n-1]->group_id) {
381 print_file(nodes[n]);
387 struct progress_state {
389 int nb_values, value;
393 void print_progress(struct progress_state *state) {
396 char buffer[PROGRESS_BUFFER_SIZE];
400 /* We use the previous bar_width to compute the position, so that
401 we avoid doing too many ioctls */
402 position = (state->bar_width * state->value) / (state->nb_values - 1);
403 if(state->bar_width <= 0 || position != state->last_position) {
404 if(!ioctl (STDERR_FILENO, TIOCGWINSZ, (char *) &win)) {
405 /* Something weird is going on if the previous test is wrong */
406 if(win.ws_col >= PROGRESS_BUFFER_SIZE - 3) {
407 state->bar_width = PROGRESS_BUFFER_SIZE - 10;
409 state->bar_width = win.ws_col - 7;
411 position = (state->bar_width * state->value) / (state->nb_values - 1);
412 state->last_position = position;
414 for(k = 0; k < position; k++) {
417 for(; k < state->bar_width; k++) {
421 /* We need four % because of the fprintf that follows */
422 sprintf(s, " [%3d%%%%]\r",
423 (100 * state->value) / (state->nb_values - 1));
425 fprintf(stderr, buffer);
431 void start(const char *dirname1, const char *dirname2) {
432 struct file_node *list1, *list2;
433 struct file_node *node1, *node2;
434 struct progress_state progress_state;
436 int nb_groups, nb_nodes;
437 int list1_length, list2_length, previous_progress;
439 char *buffer1 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
440 char *buffer2 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
445 fprintf(stderr, "Scanning %s ... ", dirname1);
448 list1 = scan_directory(0, dirname1);
450 list1_length = file_list_length(list1);
453 if(strncmp(dirname2, "not:", 4) == 0) {
455 /* groups are not computed in the not: mode */
458 } else if(strncmp(dirname2, "and:", 4) == 0) {
462 fprintf(stderr, "%s ... ", dirname2);
464 list2 = scan_directory(0, dirname2);
470 fprintf(stderr, "done.\n");
473 dirname1, list1_length, (list1_length > 1 ? "s" : ""));
475 list2_length = file_list_length(list2);
478 dirname2, list2_length, (list2_length > 1 ? "s" : ""));
480 fprintf(stderr, "Now looking for identical files.\n");
484 previous_progress = -1;
487 progress_state.bar_width = -1;
488 progress_state.last_position = -1;
489 progress_state.nb_values = list1_length;
492 for(node1 = list1; node1; node1 = node1->next) {
493 progress_state.value = nb_nodes;
494 print_progress(&progress_state);
499 for(node2 = list2; !found && node2; node2 = node2->next) {
500 if(same_files(node1, node2, buffer1, buffer2)) {
507 printf("%s\n", realpath(node1->name, 0));
509 printf("%s\n", node1->name);
515 for(node1 = list1; node1; node1 = node1->next) {
516 progress_state.value = nb_nodes;
517 print_progress(&progress_state);
520 for(node2 = list2; node2; node2 = node2->next) {
521 if(node1->group_id < 0 || node2->group_id < 0) {
522 if(same_files(node1, node2, buffer1, buffer2)) {
523 if(node1->group_id < 0) {
524 if(node2->group_id >= 0) {
525 node1->group_id = node2->group_id;
527 node1->group_id = nb_groups;
532 if(node2->group_id < 0) {
533 node2->group_id = node1->group_id;
543 fprintf(stderr, "\n");
547 print_result(list1, list2);
548 file_list_delete(list1);
549 file_list_delete(list2);
551 print_result(list1, 0);
552 file_list_delete(list1);
559 void usage(FILE *out) {
560 fprintf(out, "Usage: finddup [OPTION]... DIR1 [[and:|not:]DIR2]\n");
561 fprintf(out, "Version %s (%s)\n", VERSION_NUMBER, UNAME);
562 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");
564 /* 01234567890123456789012345678901234567890123456789012345678901234567890123456789*/
565 fprintf(out, " -h, --help show this help\n");
566 fprintf(out, " -d, --ignore-dots ignore dot files and directories\n");
567 fprintf(out, " -0, --ignore-empty ignore empty files\n");
568 fprintf(out, " -c, --hide-matchings do not show which files in DIR2 corresponds to\n");
569 fprintf(out, " those in DIR1\n");
570 fprintf(out, " -g, --no-group-ids do not show the file groups\n");
571 fprintf(out, " -p, --show-progress show progress\n");
572 fprintf(out, " -r, --real-paths show the real file paths\n");
573 fprintf(out, " -i, --same-inodes-are-different\n");
574 fprintf(out, " consider files with same inode as different\n");
576 fprintf(out, " -m, --md5 use MD5 hashing\n");
579 fprintf(out, "Report bugs and comments to <francois@fleuret.org>.\n");
582 /**********************************************************************/
584 static struct option long_options[] = {
585 { "help", no_argument, 0, 'h' },
586 { "same-inodes-are-different", no_argument, 0, 'i' },
587 { "real-paths", no_argument, 0, 'r' },
588 { "hide-matchings", no_argument, 0, 'c' },
589 { "no-group-ids", no_argument, 0, 'g' },
590 { "ignore-dots", no_argument, 0, 'd' },
591 { "ignore-empty", no_argument, 0, '0' },
592 { "show-progress", no_argument, 0, 'p' },
593 { "md5", no_argument, 0, 'm' },
597 int main(int argc, char **argv) {
600 setlocale (LC_ALL, "");
602 while ((c = getopt_long(argc, argv, "hircgd0pm",
603 long_options, NULL)) != -1) {
617 ignore_empty_files = 1;
625 same_inodes_are_different = 1;
645 "finddup has not been compiled with MD5 hashing.\n");
657 if(!isatty(STDERR_FILENO)) {
661 if(optind + 2 == argc) {
662 start(argv[optind], argv[optind + 1]);
663 } else if(optind + 1 == argc) {
664 same_inodes_are_different = 1;
665 start(argv[optind], 0);