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 "1.1"
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 /********************************************************************/
83 /* malloc with error checking. */
85 void *safe_malloc(size_t n) {
89 "finddup: Can not allocate memory: %s\n", strerror(errno));
95 /********************************************************************/
97 int ignore_entry(const char *name) {
99 strcmp(name, ".") == 0 ||
100 strcmp(name, "..") == 0 ||
101 (ignore_dotfiles && name[0] == '.' && name[1] != '/');
104 /**********************************************************************/
107 struct file_node *next;
110 time_t atime, mtime, ctime;
112 int group_id; /* one per identical file content */
113 int dir_id; /* 1 for DIR1, and 2 for DIR2 */
116 void file_list_delete(struct file_node *head) {
117 struct file_node *next;
126 int file_list_length(struct file_node *head) {
135 /**********************************************************************/
137 int same_content(struct file_node *f1, struct file_node *f2,
138 char *buffer1, char *buffer2) {
139 int fd1, fd2, s1, s2;
141 fd1 = open(f1->name, O_RDONLY);
142 fd2 = open(f2->name, O_RDONLY);
144 if(fd1 >= 0 && fd2 >= 0) {
146 s1 = read(fd1, buffer1, READ_BUFFER_SIZE);
147 s2 = read(fd2, buffer2, READ_BUFFER_SIZE);
149 if(s1 < 0 || s2 < 0) {
161 if(memcmp(buffer1, buffer2, s1)) {
162 /* printf("size_to_read = %d\n", size_to_read); */
170 "finddup: Different read size without error on files of same size.\n");
178 "finddup: Can not open \"%s\" error: %s\n",
185 "finddup: Can not open \"%s\" error: %s\n",
194 int same_files(struct file_node *f1, struct file_node *f2,
195 char *buffer1, char *buffer2) {
196 if(same_inodes_are_different && f1->inode == f2->inode) {
200 return f1->size == f2->size && same_content(f1, f2, buffer1, buffer2);
203 /**********************************************************************/
205 struct file_node *scan_directory_rec(struct file_node *tail, const char *name) {
207 struct dirent *dir_e;
209 struct file_node *tmp;
210 char subname[PATH_MAX + 1];
212 if(lstat(name, &sb) != 0) {
213 fprintf(stderr, "finddup: Can not stat \"%s\": %s\n", name, strerror(errno));
217 if(S_ISLNK(sb.st_mode)) {
224 while((dir_e = readdir(dir))) {
225 if(!ignore_entry(dir_e->d_name)) {
226 snprintf(subname, PATH_MAX, "%s/%s", name, dir_e->d_name);
227 tail = scan_directory_rec(tail, subname);
232 if(S_ISREG(sb.st_mode)) {
233 if(!ignore_entry(name)) {
234 if(!ignore_empty_files || sb.st_size > 0) {
235 tmp = safe_malloc(sizeof(struct file_node));
237 tmp->name = strdup(name);
238 tmp->size = sb.st_size;
239 tmp->atime = sb.st_atime;
240 tmp->mtime = sb.st_mtime;
241 tmp->ctime = sb.st_ctime;
242 tmp->inode = sb.st_ino;
254 struct file_node *scan_directory(struct file_node *tail, const char *name) {
255 struct file_node *result;
259 fprintf(stderr, "Scanning '%s' ... ", name);
262 result = scan_directory_rec(tail, name);
263 length = file_list_length(result);
266 fprintf(stderr, "done (%d file%s).\n",
267 length, (length > 1 ? "s" : ""));
274 void print_file(struct file_node *node) {
275 char tmp[PATH_MAX + 1];
277 if(realpath(node->name, tmp)) {
279 printf("%d %s\n", node->group_id, tmp);
285 "finddup: Can not get the realpath of \"%s\": %s\n",
292 printf("%d %s\n", node->group_id, node->name);
294 printf("%s\n", node->name);
299 int compare_nodes(const void *x1, const void *x2) {
300 const struct file_node **f1, **f2;
302 f1 = (const struct file_node **) x1;
303 f2 = (const struct file_node **) x2;
305 if((*f1)->group_id < (*f2)->group_id) {
307 } else if((*f1)->group_id > (*f2)->group_id) {
311 if((*f1)->mtime < (*f2)->mtime) {
313 } else if((*f1)->mtime > (*f2)->mtime) {
319 if((*f1)->dir_id < (*f2)->dir_id) {
321 } else if((*f1)->dir_id > (*f2)->dir_id) {
330 void exec_command(int nb, struct file_node **nodes) {
332 int max_group_size = 0, group_size, m, n, status;
335 for(n = 0; n < nb; n++) {
336 if(n == 0 || nodes[n]->group_id != nodes[n-1]->group_id) {
340 if(group_size > max_group_size) {
341 max_group_size = group_size;
345 args = safe_malloc((max_group_size + 2) * sizeof(char *));
346 args[0] = command_to_exec;
351 while(n < nb && nodes[n]->group_id == nodes[m]->group_id) {
352 args[n - m + 1] = nodes[n]->name;
358 } else if(pid == 0) {
359 if(execvp(command_to_exec, args) < 0) {
364 while(wait(&status) != pid);
374 void print_result(struct file_node *list1, struct file_node *list2) {
375 struct file_node *node1, *node2;
376 struct file_node **nodes;
380 for(node1 = list1; node1; node1 = node1->next) {
381 if(node1->group_id >= 0) { nb++; }
385 for(node2 = list2; node2; node2 = node2->next) {
386 if(node2->group_id >= 0) { nb++; }
390 nodes = safe_malloc(nb * sizeof(struct file_node *));
393 for(node1 = list1; node1; node1 = node1->next) {
394 if(node1->group_id >= 0) {
400 for(node2 = list2; node2; node2 = node2->next) {
401 if(node2->group_id >= 0) {
407 qsort(nodes, nb, sizeof(struct file_node *), compare_nodes);
409 if(command_to_exec) {
410 exec_command(nb, nodes);
412 for(n = 0; n < nb; n++) {
413 if(!show_groups && n > 0 && nodes[n]->group_id != nodes[n-1]->group_id) {
416 print_file(nodes[n]);
423 struct progress_state {
425 int nb_values, value;
429 void print_progress(struct progress_state *state) {
430 int position, k, normalizer;
432 char buffer[PROGRESS_BUFFER_SIZE];
435 normalizer = (state->nb_values > 1 ? state->nb_values - 1 : 1);
438 /* We use the previous bar_width to compute the position, so that
439 we avoid doing too many ioctls */
440 position = (state->bar_width * state->value) / normalizer;
441 if(state->bar_width <= 0 || position != state->last_position) {
442 if(!ioctl (STDERR_FILENO, TIOCGWINSZ, (char *) &win)) {
443 /* Something weird is going on if the previous test is wrong */
444 if(win.ws_col >= PROGRESS_BUFFER_SIZE - 3) {
445 state->bar_width = PROGRESS_BUFFER_SIZE - 10;
447 state->bar_width = win.ws_col - 7;
449 position = (state->bar_width * state->value) / normalizer;
450 state->last_position = position;
452 for(k = 0; k < position; k++) {
455 for(; k < state->bar_width; k++) {
459 /* We need four % because of the fprintf that follows */
460 sprintf(s, " [%3d%%%%]\r",
461 (100 * state->value) / normalizer);
463 fprintf(stderr, buffer);
469 void start(const char *dirname1, const char *dirname2) {
470 struct file_node *list1, *list2;
471 struct file_node *node1, *node2;
472 struct progress_state progress_state;
474 int nb_groups, nb_nodes;
475 int list1_length, previous_progress;
477 char *buffer1 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
478 char *buffer2 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
482 list1 = scan_directory(0, dirname1);
483 list1_length = file_list_length(list1);
486 if(strncmp(dirname2, "not:", 4) == 0) {
488 /* groups are not computed in the not: mode */
491 } else if(strncmp(dirname2, "and:", 4) == 0) {
494 list2 = scan_directory(0, dirname2);
501 "Now looking for identical files (this may take a while).\n");
505 previous_progress = -1;
508 progress_state.bar_width = -1;
509 progress_state.last_position = -1;
510 progress_state.nb_values = list1_length;
513 for(node1 = list1; node1; node1 = node1->next) {
514 progress_state.value = nb_nodes;
515 print_progress(&progress_state);
520 for(node2 = list2; !found && node2; node2 = node2->next) {
521 if(same_files(node1, node2, buffer1, buffer2)) {
528 printf("%s\n", realpath(node1->name, 0));
530 printf("%s\n", node1->name);
536 for(node1 = list1; node1; node1 = node1->next) {
537 progress_state.value = nb_nodes;
538 print_progress(&progress_state);
541 for(node2 = list2; node2; node2 = node2->next) {
542 if(node1->group_id < 0 || node2->group_id < 0) {
543 if(same_files(node1, node2, buffer1, buffer2)) {
544 if(node1->group_id < 0) {
545 if(node2->group_id >= 0) {
546 node1->group_id = node2->group_id;
548 node1->group_id = nb_groups;
553 if(node2->group_id < 0) {
554 node2->group_id = node1->group_id;
564 fprintf(stderr, "\n");
568 print_result(list1, list2);
569 file_list_delete(list1);
570 file_list_delete(list2);
572 print_result(list1, 0);
573 file_list_delete(list1);
580 void usage(FILE *out) {
581 fprintf(out, "Usage: finddup [OPTION]... [DIR1 [[and:|not:]DIR2]]\n");
582 fprintf(out, "Version %s (%s)\n", VERSION_NUMBER, UNAME);
583 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");
585 /* 01234567890123456789012345678901234567890123456789012345678901234567890123456789*/
586 fprintf(out, " -v, --version prints the version number and exit\n");
587 fprintf(out, " -h, --help show this help\n");
588 fprintf(out, " -d, --ignore-dots ignore dot files and directories\n");
589 fprintf(out, " -0, --ignore-empty ignore empty files\n");
590 fprintf(out, " -c, --hide-matchings do not show which files in DIR2 corresponds to\n");
591 fprintf(out, " those in DIR1\n");
592 fprintf(out, " -g, --no-group-ids do not show the file groups\n");
593 fprintf(out, " -t, --time-sort sort according to modification time in each group\n");
594 fprintf(out, " -p, --show-progress show progress\n");
595 fprintf(out, " -r, --real-paths show the real file paths\n");
596 fprintf(out, " -i, --same-inodes-are-different\n");
597 fprintf(out, " consider files with same inode as different\n");
598 fprintf(out, " -e <command>, --exec <command>\n");
599 fprintf(out, " execute the provided command for each group of\n");
600 fprintf(out, " identical files, with their names as arguments\n");
602 fprintf(out, "Report bugs and comments to <francois@fleuret.org>.\n");
605 /**********************************************************************/
607 static struct option long_options[] = {
608 { "version", no_argument, 0, 'v' },
609 { "help", no_argument, 0, 'h' },
610 { "same-inodes-are-different", no_argument, 0, 'i' },
611 { "real-paths", no_argument, 0, 'r' },
612 { "hide-matchings", no_argument, 0, 'c' },
613 { "no-group-ids", no_argument, 0, 'g' },
614 { "time-sort", no_argument, 0, 't' },
615 { "ignore-dots", no_argument, 0, 'd' },
616 { "ignore-empty", no_argument, 0, '0' },
617 { "show-progress", no_argument, 0, 'p' },
618 { "exec", 1, 0, 'e' },
622 int main(int argc, char **argv) {
625 setlocale (LC_ALL, "");
627 while ((c = getopt_long(argc, argv, "vhircgtd0pme:",
628 long_options, NULL)) != -1) {
632 printf("finddup version %s (%s)\n", VERSION_NUMBER, UNAME);
647 ignore_empty_files = 1;
655 same_inodes_are_different = 1;
675 if(command_to_exec != 0) {
676 free(command_to_exec);
678 command_to_exec = strdup(optarg);
687 if(!isatty(STDERR_FILENO)) {
691 if(optind + 2 == argc) {
692 start(argv[optind], argv[optind + 1]);
693 } else if(optind + 1 == argc) {
694 same_inodes_are_different = 1;
695 start(argv[optind], 0);
696 } else if(optind == argc) {
697 same_inodes_are_different = 1;