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 int trim_first = 0; /* remove the first entry in each group */
80 char *command_to_exec = 0; /* the name of the command to exec for each
81 group of identical files */
83 char *result_file_prefix = 0; /* The prefix to use to write result */
85 /********************************************************************/
87 /* malloc with error checking. */
89 void *safe_malloc(size_t n) {
93 "finddup: Can not allocate memory: %s\n", strerror(errno));
99 /********************************************************************/
101 int ignore_entry(const char *name) {
103 strcmp(name, ".") == 0 ||
104 strcmp(name, "..") == 0 ||
105 (ignore_dotfiles && name[0] == '.' && name[1] != '/');
108 /**********************************************************************/
111 struct file_node *next;
114 time_t atime, mtime, ctime;
116 int group_id; /* one per identical file content */
117 int dir_id; /* 1 for DIR1, and 2 for DIR2 */
120 void file_list_delete(struct file_node *head) {
121 struct file_node *next;
130 int file_list_length(struct file_node *head) {
139 /**********************************************************************/
141 int same_content(struct file_node *f1, struct file_node *f2,
142 char *buffer1, char *buffer2) {
143 int fd1, fd2, s1, s2;
145 fd1 = open(f1->name, O_RDONLY);
146 fd2 = open(f2->name, O_RDONLY);
148 if(fd1 >= 0 && fd2 >= 0) {
150 s1 = read(fd1, buffer1, READ_BUFFER_SIZE);
151 s2 = read(fd2, buffer2, READ_BUFFER_SIZE);
153 if(s1 < 0 || s2 < 0) {
165 if(memcmp(buffer1, buffer2, s1)) {
166 /* printf("size_to_read = %d\n", size_to_read); */
174 "finddup: Different read size without error on files of same size.\n");
182 "finddup: Can not open \"%s\" error: %s\n",
189 "finddup: Can not open \"%s\" error: %s\n",
198 int same_files(struct file_node *f1, struct file_node *f2,
199 char *buffer1, char *buffer2) {
200 if(same_inodes_are_different && f1->inode == f2->inode) {
204 return f1->size == f2->size && same_content(f1, f2, buffer1, buffer2);
207 /**********************************************************************/
209 struct file_node *scan_directory_rec(struct file_node *tail, const char *name) {
211 struct dirent *dir_e;
213 struct file_node *tmp;
214 char subname[PATH_MAX + 1];
216 if(lstat(name, &sb) != 0) {
217 fprintf(stderr, "finddup: Can not stat \"%s\": %s\n", name, strerror(errno));
221 if(S_ISLNK(sb.st_mode)) {
228 while((dir_e = readdir(dir))) {
229 if(!ignore_entry(dir_e->d_name)) {
230 snprintf(subname, PATH_MAX, "%s/%s", name, dir_e->d_name);
231 tail = scan_directory_rec(tail, subname);
236 if(S_ISREG(sb.st_mode)) {
237 if(!ignore_entry(name)) {
238 if(!ignore_empty_files || sb.st_size > 0) {
239 tmp = safe_malloc(sizeof(struct file_node));
241 tmp->name = strdup(name);
242 tmp->size = sb.st_size;
243 tmp->atime = sb.st_atime;
244 tmp->mtime = sb.st_mtime;
245 tmp->ctime = sb.st_ctime;
246 tmp->inode = sb.st_ino;
258 struct file_node *scan_directory(struct file_node *tail, const char *name) {
259 struct file_node *result;
263 fprintf(stderr, "Scanning '%s' ... ", name);
266 result = scan_directory_rec(tail, name);
267 length = file_list_length(result);
270 fprintf(stderr, "done (%d file%s).\n",
271 length, (length > 1 ? "s" : ""));
278 void write_one_entry_to_file(FILE *out, struct file_node *node) {
279 char tmp[PATH_MAX + 1];
281 if(realpath(node->name, tmp)) {
283 fprintf(out, "%d %s\n", node->group_id, tmp);
285 fprintf(out, "%s\n", tmp);
289 "finddup: Can not get the realpath of \"%s\": %s\n",
296 fprintf(out, "%d %s\n", node->group_id, node->name);
298 fprintf(out, "%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) {
315 if((*f1)->mtime < (*f2)->mtime) {
317 } else if((*f1)->mtime > (*f2)->mtime) {
323 if((*f1)->dir_id < (*f2)->dir_id) {
325 } else if((*f1)->dir_id > (*f2)->dir_id) {
334 void exec_command(int nb, struct file_node **nodes) {
336 int max_group_size = 0, group_size = 0, m, n, status;
339 for(n = 0; n < nb; n++) {
340 if(nodes[n]->group_id != nodes[n-1]->group_id) {
344 if(group_size > max_group_size) {
345 max_group_size = group_size;
349 args = safe_malloc((max_group_size + 2) * sizeof(char *));
350 args[0] = command_to_exec;
355 if(trim_first) { m++; }
356 while(n < nb && nodes[n]->group_id == nodes[m]->group_id) {
358 args[n - m + 1] = nodes[n]->name;
365 } else if(pid == 0) {
366 if(execvp(command_to_exec, args) < 0) {
371 while(wait(&status) != pid);
381 void write_groups_in_files(int nb, struct file_node **nodes) {
383 int current_group = -1, n, first_of_group;
384 char filename[PATH_MAX + 1];
386 for(n = 0; n < nb; n++) {
387 first_of_group = (n == 0);
388 if(nodes[n]->group_id != current_group) {
389 if(file) { fclose(file); }
390 sprintf(filename, "%s%06d", result_file_prefix, nodes[n]->group_id);
391 file = fopen(filename, "w");
392 current_group = nodes[n]->group_id;
393 printf("Writing %s.\n" , filename);
397 if(!trim_first || !first_of_group) {
398 write_one_entry_to_file(file, nodes[n]);
402 if(file) { fclose(file); }
405 void print_result(struct file_node *list1, struct file_node *list2) {
406 struct file_node *node1, *node2;
407 struct file_node **nodes;
408 int nb, n, first_of_group;
411 for(node1 = list1; node1; node1 = node1->next) {
412 if(node1->group_id >= 0) { nb++; }
416 for(node2 = list2; node2; node2 = node2->next) {
417 if(node2->group_id >= 0) { nb++; }
421 nodes = safe_malloc(nb * sizeof(struct file_node *));
424 for(node1 = list1; node1; node1 = node1->next) {
425 if(node1->group_id >= 0) {
431 for(node2 = list2; node2; node2 = node2->next) {
432 if(node2->group_id >= 0) {
438 qsort(nodes, nb, sizeof(struct file_node *), compare_nodes);
440 if(command_to_exec) {
441 exec_command(nb, nodes);
442 } else if(result_file_prefix) {
443 write_groups_in_files(nb, nodes);
445 for(n = 0; n < nb; n++) {
446 first_of_group = (n == 0);
447 if(n > 0 && nodes[n]->group_id != nodes[n-1]->group_id) {
453 if(!trim_first || !first_of_group) {
454 write_one_entry_to_file(stdout, nodes[n]);
462 struct progress_state {
464 int nb_values, value;
468 void print_progress(struct progress_state *state) {
469 int position, k, normalizer;
471 char buffer[PROGRESS_BUFFER_SIZE];
474 normalizer = (state->nb_values > 1 ? state->nb_values - 1 : 1);
477 /* We use the previous bar_width to compute the position, so that
478 we avoid doing too many ioctls */
479 position = (state->bar_width * state->value) / normalizer;
480 if(state->bar_width <= 0 || position != state->last_position) {
481 if(!ioctl (STDERR_FILENO, TIOCGWINSZ, (char *) &win)) {
482 /* Something weird is going on if the previous test is wrong */
483 if(win.ws_col >= PROGRESS_BUFFER_SIZE - 3) {
484 state->bar_width = PROGRESS_BUFFER_SIZE - 10;
486 state->bar_width = win.ws_col - 7;
488 position = (state->bar_width * state->value) / normalizer;
489 state->last_position = position;
491 for(k = 0; k < position; k++) {
494 for(; k < state->bar_width; k++) {
498 /* We need four % because of the fprintf that follows */
499 sprintf(s, " [%3d%%]\r",
500 (100 * state->value) / normalizer);
502 fprintf(stderr, "%s", buffer);
508 void start(const char *dirname1, const char *dirname2) {
509 struct file_node *list1, *list2;
510 struct file_node *node1, *node2;
511 struct progress_state progress_state;
513 int nb_groups, nb_nodes;
516 char *buffer1 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
517 char *buffer2 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
521 list1 = scan_directory(0, dirname1);
522 list1_length = file_list_length(list1);
525 if(strncmp(dirname2, "not:", 4) == 0) {
527 /* groups are not computed in the not: mode */
530 } else if(strncmp(dirname2, "and:", 4) == 0) {
533 list2 = scan_directory(0, dirname2);
540 "Now looking for identical files (this may take a while).\n");
546 progress_state.bar_width = -1;
547 progress_state.last_position = -1;
548 progress_state.nb_values = list1_length;
551 for(node1 = list1; node1; node1 = node1->next) {
552 progress_state.value = nb_nodes;
553 print_progress(&progress_state);
558 for(node2 = list2; !found && node2; node2 = node2->next) {
559 if(same_files(node1, node2, buffer1, buffer2)) {
566 printf("%s\n", realpath(node1->name, 0));
568 printf("%s\n", node1->name);
574 for(node1 = list1; node1; node1 = node1->next) {
575 progress_state.value = nb_nodes;
576 print_progress(&progress_state);
579 for(node2 = list2; node2; node2 = node2->next) {
580 if(node1->group_id < 0 || node2->group_id < 0) {
581 if(same_files(node1, node2, buffer1, buffer2)) {
582 if(node1->group_id < 0) {
583 if(node2->group_id >= 0) {
584 node1->group_id = node2->group_id;
586 node1->group_id = nb_groups;
591 if(node2->group_id < 0) {
592 node2->group_id = node1->group_id;
602 fprintf(stderr, "\n");
606 print_result(list1, list2);
607 file_list_delete(list1);
608 file_list_delete(list2);
610 print_result(list1, 0);
611 file_list_delete(list1);
618 void usage(FILE *out) {
619 fprintf(out, "Usage: finddup [OPTION]... [DIR1 [[and:|not:]DIR2]]\n");
620 fprintf(out, "Version %s (%s)\n", VERSION_NUMBER, UNAME);
621 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");
623 /* 01234567890123456789012345678901234567890123456789012345678901234567890123456789*/
624 fprintf(out, " -v, --version prints the version number and exit\n");
625 fprintf(out, " -h, --help show this help\n");
626 fprintf(out, " -d, --ignore-dots ignore dot files and directories\n");
627 fprintf(out, " -0, --ignore-empty ignore empty files\n");
628 fprintf(out, " -c, --hide-matchings do not show which files in DIR2 corresponds to\n");
629 fprintf(out, " those in DIR1\n");
630 fprintf(out, " -g, --no-group-ids do not show the file groups\n");
631 fprintf(out, " -t, --time-sort sort according to modification time in each group\n");
632 fprintf(out, " -q, --trim-first do not show the first file in each group\n");
633 fprintf(out, " -p, --show-progress show progress\n");
634 fprintf(out, " -r, --real-paths show the real file paths\n");
635 fprintf(out, " -i, --same-inodes-are-different\n");
636 fprintf(out, " consider files with same inode as different\n");
637 fprintf(out, " -e <command>, --exec <command>\n");
638 fprintf(out, " execute the provided command for each group of\n");
639 fprintf(out, " identical files, with their names as arguments\n");
640 fprintf(out, " -f <string>, --result-prefix <string>\n");
641 fprintf(out, " for each group of identical files, write one\n");
642 fprintf(out, " result file whose name is the given prefix string\n");
643 fprintf(out, " followed by the group number, and containing\n");
644 fprintf(out, " one filename per line\n");
646 fprintf(out, "Report bugs and comments to <francois@fleuret.org>.\n");
649 /**********************************************************************/
651 static struct option long_options[] = {
652 { "version", no_argument, 0, 'v' },
653 { "help", no_argument, 0, 'h' },
654 { "same-inodes-are-different", no_argument, 0, 'i' },
655 { "real-paths", no_argument, 0, 'r' },
656 { "hide-matchings", no_argument, 0, 'c' },
657 { "no-group-ids", no_argument, 0, 'g' },
658 { "time-sort", no_argument, 0, 't' },
659 { "trim-first", no_argument, 0, 'q' },
660 { "ignore-dots", no_argument, 0, 'd' },
661 { "ignore-empty", no_argument, 0, '0' },
662 { "show-progress", no_argument, 0, 'p' },
663 { "exec", 1, 0, 'e' },
664 { "result-prefix", 1, 0, 'f' },
668 int main(int argc, char **argv) {
671 setlocale (LC_ALL, "");
673 while ((c = getopt_long(argc, argv, "vhircgtqd0pme:f:",
674 long_options, NULL)) != -1) {
678 printf("finddup version %s (%s)\n", VERSION_NUMBER, UNAME);
693 ignore_empty_files = 1;
701 same_inodes_are_different = 1;
725 if(command_to_exec != 0) {
726 free(command_to_exec);
728 command_to_exec = strdup(optarg);
732 if(result_file_prefix != 0) {
733 free(result_file_prefix);
735 result_file_prefix = strdup(optarg);
745 if(!isatty(STDERR_FILENO)) {
749 if(optind + 2 == argc) {
750 start(argv[optind], argv[optind + 1]);
751 } else if(optind + 1 == argc) {
752 same_inodes_are_different = 1;
753 start(argv[optind], 0);
754 } else if(optind == argc) {
755 same_inodes_are_different = 1;