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, 2011 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 "1.2"
30 #include <sys/types.h>
32 #include <sys/param.h>
40 #include <sys/ioctl.h>
45 /* 1M really helps compared to 64k */
46 #define READ_BUFFER_SIZE (1024 * 1024)
48 #define PROGRESS_BUFFER_SIZE 1024
50 typedef int64_t size_sum_t;
52 /* Yeah, global variables! */
54 int ignore_dotfiles = 0; /* 1 means ignore files and directories
55 starting with a dot */
57 int ignore_empty_files = 0; /* 1 means ignore empty files */
59 int show_realpaths = 0; /* 1 means ignore files and directories
60 starting with a dot */
62 int show_progress = 0; /* 1 means show a progress bar when we are in a
65 int show_hits = 1; /* 1 means to show the files from dir2
66 corresponding to the ones from dir1 */
68 int show_groups = 1; /* 1 means to show the group IDs when printing
71 int same_inodes_are_different = 0; /* 1 means that comparison between
72 two files with same inode will
75 int sort_by_time = 0; /* 1 means to sort files in each group according
76 to the modification time */
78 char *command_to_exec = 0; /* the name of the command to exec for each
79 group of identical files */
81 char *result_file_prefix = 0; /* The prefix to use to write result */
83 /********************************************************************/
85 /* malloc with error checking. */
87 void *safe_malloc(size_t n) {
91 "finddup: Can not allocate memory: %s\n", strerror(errno));
97 /********************************************************************/
99 int ignore_entry(const char *name) {
101 strcmp(name, ".") == 0 ||
102 strcmp(name, "..") == 0 ||
103 (ignore_dotfiles && name[0] == '.' && name[1] != '/');
106 /**********************************************************************/
109 struct file_node *next;
112 time_t atime, mtime, ctime;
114 int group_id; /* one per identical file content */
115 int dir_id; /* 1 for DIR1, and 2 for DIR2 */
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;
143 fd1 = open(f1->name, O_RDONLY);
144 fd2 = open(f2->name, O_RDONLY);
146 if(fd1 >= 0 && fd2 >= 0) {
148 s1 = read(fd1, buffer1, READ_BUFFER_SIZE);
149 s2 = read(fd2, buffer2, READ_BUFFER_SIZE);
151 if(s1 < 0 || s2 < 0) {
163 if(memcmp(buffer1, buffer2, s1)) {
164 /* printf("size_to_read = %d\n", size_to_read); */
172 "finddup: Different read size without error on files of same size.\n");
180 "finddup: Can not open \"%s\" error: %s\n",
187 "finddup: Can not open \"%s\" error: %s\n",
196 int same_files(struct file_node *f1, struct file_node *f2,
197 char *buffer1, char *buffer2) {
198 if(same_inodes_are_different && f1->inode == f2->inode) {
202 return f1->size == f2->size && same_content(f1, f2, buffer1, buffer2);
205 /**********************************************************************/
207 struct file_node *scan_directory_rec(struct file_node *tail, const char *name) {
209 struct dirent *dir_e;
211 struct file_node *tmp;
212 char subname[PATH_MAX + 1];
214 if(lstat(name, &sb) != 0) {
215 fprintf(stderr, "finddup: Can not stat \"%s\": %s\n", name, strerror(errno));
219 if(S_ISLNK(sb.st_mode)) {
226 while((dir_e = readdir(dir))) {
227 if(!ignore_entry(dir_e->d_name)) {
228 snprintf(subname, PATH_MAX, "%s/%s", name, dir_e->d_name);
229 tail = scan_directory_rec(tail, subname);
234 if(S_ISREG(sb.st_mode)) {
235 if(!ignore_entry(name)) {
236 if(!ignore_empty_files || sb.st_size > 0) {
237 tmp = safe_malloc(sizeof(struct file_node));
239 tmp->name = strdup(name);
240 tmp->size = sb.st_size;
241 tmp->atime = sb.st_atime;
242 tmp->mtime = sb.st_mtime;
243 tmp->ctime = sb.st_ctime;
244 tmp->inode = sb.st_ino;
256 struct file_node *scan_directory(struct file_node *tail, const char *name) {
257 struct file_node *result;
261 fprintf(stderr, "Scanning '%s' ... ", name);
264 result = scan_directory_rec(tail, name);
265 length = file_list_length(result);
268 fprintf(stderr, "done (%d file%s).\n",
269 length, (length > 1 ? "s" : ""));
276 void print_file(FILE *out, struct file_node *node) {
277 char tmp[PATH_MAX + 1];
279 if(realpath(node->name, tmp)) {
281 fprintf(out, "%d %s\n", node->group_id, tmp);
283 fprintf(out, "%s\n", tmp);
287 "finddup: Can not get the realpath of \"%s\": %s\n",
294 fprintf(out, "%d %s\n", node->group_id, node->name);
296 fprintf(out, "%s\n", node->name);
301 int compare_nodes(const void *x1, const void *x2) {
302 const struct file_node **f1, **f2;
304 f1 = (const struct file_node **) x1;
305 f2 = (const struct file_node **) x2;
307 if((*f1)->group_id < (*f2)->group_id) {
309 } else if((*f1)->group_id > (*f2)->group_id) {
313 if((*f1)->mtime < (*f2)->mtime) {
315 } else if((*f1)->mtime > (*f2)->mtime) {
321 if((*f1)->dir_id < (*f2)->dir_id) {
323 } else if((*f1)->dir_id > (*f2)->dir_id) {
332 void exec_command(int nb, struct file_node **nodes) {
334 int max_group_size = 0, group_size, m, n, status;
337 for(n = 0; n < nb; n++) {
338 if(n == 0 || nodes[n]->group_id != nodes[n-1]->group_id) {
342 if(group_size > max_group_size) {
343 max_group_size = group_size;
347 args = safe_malloc((max_group_size + 2) * sizeof(char *));
348 args[0] = command_to_exec;
353 while(n < nb && nodes[n]->group_id == nodes[m]->group_id) {
354 args[n - m + 1] = nodes[n]->name;
360 } else if(pid == 0) {
361 if(execvp(command_to_exec, args) < 0) {
366 while(wait(&status) != pid);
376 void write_groups_in_files(int nb, struct file_node **nodes) {
378 int current_group = -1, n;
379 char filename[PATH_MAX + 1];
381 for(n = 0; n < nb; n++) {
382 if(nodes[n]->group_id != current_group) {
383 if(file) { fclose(file); }
384 sprintf(filename, "%s%06d", result_file_prefix, nodes[n]->group_id);
385 file = fopen(filename, "w");
386 current_group = nodes[n]->group_id;
387 printf("Writing %s.\n" , filename);
389 print_file(file, nodes[n]);
392 if(file) { fclose(file); }
395 void print_result(struct file_node *list1, struct file_node *list2) {
396 struct file_node *node1, *node2;
397 struct file_node **nodes;
401 for(node1 = list1; node1; node1 = node1->next) {
402 if(node1->group_id >= 0) { nb++; }
406 for(node2 = list2; node2; node2 = node2->next) {
407 if(node2->group_id >= 0) { nb++; }
411 nodes = safe_malloc(nb * sizeof(struct file_node *));
414 for(node1 = list1; node1; node1 = node1->next) {
415 if(node1->group_id >= 0) {
421 for(node2 = list2; node2; node2 = node2->next) {
422 if(node2->group_id >= 0) {
428 qsort(nodes, nb, sizeof(struct file_node *), compare_nodes);
430 if(command_to_exec) {
431 exec_command(nb, nodes);
432 } else if(result_file_prefix) {
433 write_groups_in_files(nb, nodes);
435 for(n = 0; n < nb; n++) {
436 if(!show_groups && n > 0 && nodes[n]->group_id != nodes[n-1]->group_id) {
439 print_file(stdout, nodes[n]);
446 struct progress_state {
448 int nb_values, value;
452 void print_progress(struct progress_state *state) {
453 int position, k, normalizer;
455 char buffer[PROGRESS_BUFFER_SIZE];
458 normalizer = (state->nb_values > 1 ? state->nb_values - 1 : 1);
461 /* We use the previous bar_width to compute the position, so that
462 we avoid doing too many ioctls */
463 position = (state->bar_width * state->value) / normalizer;
464 if(state->bar_width <= 0 || position != state->last_position) {
465 if(!ioctl (STDERR_FILENO, TIOCGWINSZ, (char *) &win)) {
466 /* Something weird is going on if the previous test is wrong */
467 if(win.ws_col >= PROGRESS_BUFFER_SIZE - 3) {
468 state->bar_width = PROGRESS_BUFFER_SIZE - 10;
470 state->bar_width = win.ws_col - 7;
472 position = (state->bar_width * state->value) / normalizer;
473 state->last_position = position;
475 for(k = 0; k < position; k++) {
478 for(; k < state->bar_width; k++) {
482 /* We need four % because of the fprintf that follows */
483 sprintf(s, " [%3d%%]\r",
484 (100 * state->value) / normalizer);
486 fprintf(stderr, "%s", buffer);
492 void start(const char *dirname1, const char *dirname2) {
493 struct file_node *list1, *list2;
494 struct file_node *node1, *node2;
495 struct progress_state progress_state;
497 int nb_groups, nb_nodes;
498 int list1_length, previous_progress;
500 char *buffer1 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
501 char *buffer2 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
505 list1 = scan_directory(0, dirname1);
506 list1_length = file_list_length(list1);
509 if(strncmp(dirname2, "not:", 4) == 0) {
511 /* groups are not computed in the not: mode */
514 } else if(strncmp(dirname2, "and:", 4) == 0) {
517 list2 = scan_directory(0, dirname2);
524 "Now looking for identical files (this may take a while).\n");
528 previous_progress = -1;
531 progress_state.bar_width = -1;
532 progress_state.last_position = -1;
533 progress_state.nb_values = list1_length;
536 for(node1 = list1; node1; node1 = node1->next) {
537 progress_state.value = nb_nodes;
538 print_progress(&progress_state);
543 for(node2 = list2; !found && node2; node2 = node2->next) {
544 if(same_files(node1, node2, buffer1, buffer2)) {
551 printf("%s\n", realpath(node1->name, 0));
553 printf("%s\n", node1->name);
559 for(node1 = list1; node1; node1 = node1->next) {
560 progress_state.value = nb_nodes;
561 print_progress(&progress_state);
564 for(node2 = list2; node2; node2 = node2->next) {
565 if(node1->group_id < 0 || node2->group_id < 0) {
566 if(same_files(node1, node2, buffer1, buffer2)) {
567 if(node1->group_id < 0) {
568 if(node2->group_id >= 0) {
569 node1->group_id = node2->group_id;
571 node1->group_id = nb_groups;
576 if(node2->group_id < 0) {
577 node2->group_id = node1->group_id;
587 fprintf(stderr, "\n");
591 print_result(list1, list2);
592 file_list_delete(list1);
593 file_list_delete(list2);
595 print_result(list1, 0);
596 file_list_delete(list1);
603 void usage(FILE *out) {
604 fprintf(out, "Usage: finddup [OPTION]... [DIR1 [[and:|not:]DIR2]]\n");
605 fprintf(out, "Version %s (%s)\n", VERSION_NUMBER, UNAME);
606 fprintf(out, "Without DIR2, lists duplicated files found in DIR1, or the current directory if DIR1 is not provided. 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");
608 /* 01234567890123456789012345678901234567890123456789012345678901234567890123456789*/
609 fprintf(out, " -v, --version prints the version number and exit\n");
610 fprintf(out, " -h, --help show this help\n");
611 fprintf(out, " -d, --ignore-dots ignore dot files and directories\n");
612 fprintf(out, " -0, --ignore-empty ignore empty files\n");
613 fprintf(out, " -c, --hide-matchings do not show which files in DIR2 corresponds to\n");
614 fprintf(out, " those in DIR1\n");
615 fprintf(out, " -g, --no-group-ids do not show the file groups\n");
616 fprintf(out, " -t, --time-sort sort according to modification time in each group\n");
617 fprintf(out, " -p, --show-progress show progress\n");
618 fprintf(out, " -r, --real-paths show the real file paths\n");
619 fprintf(out, " -i, --same-inodes-are-different\n");
620 fprintf(out, " consider files with same inode as different\n");
621 fprintf(out, " -e <command>, --exec <command>\n");
622 fprintf(out, " execute the provided command for each group of\n");
623 fprintf(out, " identical files, with their names as arguments\n");
624 fprintf(out, " -f <string>, --result-prefix <string>\n");
625 fprintf(out, " for each group of identical files, write one\n");
626 fprintf(out, " result file whose name is the given prefix string\n");
627 fprintf(out, " followed the group number, and containing one\n");
628 fprintf(out, " filename per line\n");
630 fprintf(out, "Report bugs and comments to <francois@fleuret.org>.\n");
633 /**********************************************************************/
635 static struct option long_options[] = {
636 { "version", no_argument, 0, 'v' },
637 { "help", no_argument, 0, 'h' },
638 { "same-inodes-are-different", no_argument, 0, 'i' },
639 { "real-paths", no_argument, 0, 'r' },
640 { "hide-matchings", no_argument, 0, 'c' },
641 { "no-group-ids", no_argument, 0, 'g' },
642 { "time-sort", no_argument, 0, 't' },
643 { "ignore-dots", no_argument, 0, 'd' },
644 { "ignore-empty", no_argument, 0, '0' },
645 { "show-progress", no_argument, 0, 'p' },
646 { "exec", 1, 0, 'e' },
647 { "result-prefix", 1, 0, 'f' },
651 int main(int argc, char **argv) {
654 setlocale (LC_ALL, "");
656 while ((c = getopt_long(argc, argv, "vhircgtd0pme:f:",
657 long_options, NULL)) != -1) {
661 printf("finddup version %s (%s)\n", VERSION_NUMBER, UNAME);
676 ignore_empty_files = 1;
684 same_inodes_are_different = 1;
704 if(command_to_exec != 0) {
705 free(command_to_exec);
707 command_to_exec = strdup(optarg);
711 if(result_file_prefix != 0) {
712 free(result_file_prefix);
714 result_file_prefix = strdup(optarg);
724 if(!isatty(STDERR_FILENO)) {
728 if(optind + 2 == argc) {
729 start(argv[optind], argv[optind + 1]);
730 } else if(optind + 1 == argc) {
731 same_inodes_are_different = 1;
732 start(argv[optind], 0);
733 } else if(optind == argc) {
734 same_inodes_are_different = 1;