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>
39 #include <sys/ioctl.h>
44 /* 1M really helps compared to 64k */
45 #define READ_BUFFER_SIZE (1024 * 1024)
47 #define PROGRESS_BUFFER_SIZE 1024
49 typedef int64_t size_sum_t;
51 /* Yeah, global variables! */
53 int ignore_dotfiles = 0; /* 1 means ignore files and directories
54 starting with a dot */
56 int ignore_empty_files = 0; /* 1 means ignore empty files */
58 int show_realpaths = 0; /* 1 means ignore files and directories
59 starting with a dot */
61 int show_progress = 0; /* 1 means show a progress bar when we are in a
64 int show_hits = 1; /* 1 means to show the files from dir2
65 corresponding to the ones from dir1 */
67 int show_groups = 1; /* 1 means to show the group IDs when printing
70 int same_inodes_are_different = 0; /* 1 means that comparison between
71 two files with same inode will
74 int sort_by_time = 0; /* 1 means to sort files in each group according
75 to the modification time */
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;
105 time_t atime, mtime, ctime;
107 int group_id; /* one per identical file content */
108 int dir_id; /* 1 for DIR1, and 2 for DIR2 */
111 void file_list_delete(struct file_node *head) {
112 struct file_node *next;
121 int file_list_length(struct file_node *head) {
130 /**********************************************************************/
132 int same_content(struct file_node *f1, struct file_node *f2,
133 char *buffer1, char *buffer2) {
134 int fd1, fd2, s1, s2;
136 fd1 = open(f1->name, O_RDONLY);
137 fd2 = open(f2->name, O_RDONLY);
139 if(fd1 >= 0 && fd2 >= 0) {
141 s1 = read(fd1, buffer1, READ_BUFFER_SIZE);
142 s2 = read(fd2, buffer2, READ_BUFFER_SIZE);
144 if(s1 < 0 || s2 < 0) {
156 if(memcmp(buffer1, buffer2, s1)) {
157 /* printf("size_to_read = %d\n", size_to_read); */
165 "Different read size without error on files of same size.\n");
173 "Can not open \"%s\" error: %s\n",
180 "Can not open \"%s\" error: %s\n",
189 int same_files(struct file_node *f1, struct file_node *f2,
190 char *buffer1, char *buffer2) {
191 if(same_inodes_are_different && f1->inode == f2->inode) {
195 return f1->size == f2->size && same_content(f1, f2, buffer1, buffer2);
198 /**********************************************************************/
200 struct file_node *scan_directory(struct file_node *tail, const char *name) {
202 struct dirent *dir_e;
204 struct file_node *tmp;
205 char subname[PATH_MAX + 1];
207 if(lstat(name, &sb) != 0) {
208 fprintf(stderr, "Can not stat \"%s\": %s\n", name, strerror(errno));
212 if(S_ISLNK(sb.st_mode)) {
219 while((dir_e = readdir(dir))) {
220 if(!ignore_entry(dir_e->d_name)) {
221 snprintf(subname, PATH_MAX, "%s/%s", name, dir_e->d_name);
222 tail = scan_directory(tail, subname);
227 if(S_ISREG(sb.st_mode)) {
228 if(!ignore_entry(name)) {
229 if(!ignore_empty_files || sb.st_size > 0) {
230 tmp = safe_malloc(sizeof(struct file_node));
232 tmp->name = strdup(name);
233 tmp->size = sb.st_size;
234 tmp->atime = sb.st_atime;
235 tmp->mtime = sb.st_mtime;
236 tmp->ctime = sb.st_ctime;
237 tmp->inode = sb.st_ino;
249 void print_file(struct file_node *node) {
250 char tmp[PATH_MAX + 1];
252 if(realpath(node->name, tmp)) {
254 printf("%d %s\n", node->group_id, tmp);
259 printf("Can not get the realpath of \"%s\": %s\n",
266 printf("%d %s\n", node->group_id, node->name);
268 printf("%s\n", node->name);
273 int compare_nodes(const void *x1, const void *x2) {
274 const struct file_node **f1, **f2;
276 f1 = (const struct file_node **) x1;
277 f2 = (const struct file_node **) x2;
279 if((*f1)->group_id < (*f2)->group_id) {
281 } else if((*f1)->group_id > (*f2)->group_id) {
285 if((*f1)->mtime < (*f2)->mtime) {
287 } else if((*f1)->mtime > (*f2)->mtime) {
293 if((*f1)->dir_id < (*f2)->dir_id) {
295 } else if((*f1)->dir_id > (*f2)->dir_id) {
305 void print_result(struct file_node *list1, struct file_node *list2) {
306 struct file_node *node1, *node2;
307 struct file_node **nodes;
311 for(node1 = list1; node1; node1 = node1->next) {
312 if(node1->group_id >= 0) { nb++; }
316 for(node2 = list2; node2; node2 = node2->next) {
317 if(node2->group_id >= 0) { nb++; }
321 nodes = safe_malloc(nb * sizeof(struct file_node *));
324 for(node1 = list1; node1; node1 = node1->next) {
325 if(node1->group_id >= 0) {
331 for(node2 = list2; node2; node2 = node2->next) {
332 if(node2->group_id >= 0) {
338 qsort(nodes, nb, sizeof(struct file_node *), compare_nodes);
340 for(n = 0; n < nb; n++) {
341 if(!show_groups && n > 0 && nodes[n]->group_id != nodes[n-1]->group_id) {
344 print_file(nodes[n]);
350 struct progress_state {
352 int nb_values, value;
356 void print_progress(struct progress_state *state) {
359 char buffer[PROGRESS_BUFFER_SIZE];
363 /* We use the previous bar_width to compute the position, so that
364 we avoid doing too many ioctls */
365 position = (state->bar_width * state->value) / (state->nb_values - 1);
366 if(state->bar_width <= 0 || position != state->last_position) {
367 if(!ioctl (STDERR_FILENO, TIOCGWINSZ, (char *) &win)) {
368 /* Something weird is going on if the previous test is wrong */
369 if(win.ws_col >= PROGRESS_BUFFER_SIZE - 3) {
370 state->bar_width = PROGRESS_BUFFER_SIZE - 10;
372 state->bar_width = win.ws_col - 7;
374 position = (state->bar_width * state->value) / (state->nb_values - 1);
375 state->last_position = position;
377 for(k = 0; k < position; k++) {
380 for(; k < state->bar_width; k++) {
384 /* We need four % because of the fprintf that follows */
385 sprintf(s, " [%3d%%%%]\r",
386 (100 * state->value) / (state->nb_values - 1));
388 fprintf(stderr, buffer);
394 void start(const char *dirname1, const char *dirname2) {
395 struct file_node *list1, *list2;
396 struct file_node *node1, *node2;
397 struct progress_state progress_state;
399 int nb_groups, nb_nodes;
400 int list1_length, list2_length, previous_progress;
402 char *buffer1 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
403 char *buffer2 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
408 fprintf(stderr, "Scanning '%s' ... ", dirname1);
411 list1 = scan_directory(0, dirname1);
413 list1_length = file_list_length(list1);
416 fprintf(stderr, "done (%d file%s).\n",
417 list1_length, (list1_length > 1 ? "s" : ""));
421 if(strncmp(dirname2, "not:", 4) == 0) {
423 /* groups are not computed in the not: mode */
426 } else if(strncmp(dirname2, "and:", 4) == 0) {
430 fprintf(stderr, "Scanning '%s' ... ", dirname2);
432 list2 = scan_directory(0, dirname2);
434 list2_length = file_list_length(list2);
435 fprintf(stderr, "done (%d file%s).\n",
436 list2_length, (list2_length > 1 ? "s" : ""));
443 fprintf(stderr, "Now looking for identical files (this may take a while).\n");
447 previous_progress = -1;
450 progress_state.bar_width = -1;
451 progress_state.last_position = -1;
452 progress_state.nb_values = list1_length;
455 for(node1 = list1; node1; node1 = node1->next) {
456 progress_state.value = nb_nodes;
457 print_progress(&progress_state);
462 for(node2 = list2; !found && node2; node2 = node2->next) {
463 if(same_files(node1, node2, buffer1, buffer2)) {
470 printf("%s\n", realpath(node1->name, 0));
472 printf("%s\n", node1->name);
478 for(node1 = list1; node1; node1 = node1->next) {
479 progress_state.value = nb_nodes;
480 print_progress(&progress_state);
483 for(node2 = list2; node2; node2 = node2->next) {
484 if(node1->group_id < 0 || node2->group_id < 0) {
485 if(same_files(node1, node2, buffer1, buffer2)) {
486 if(node1->group_id < 0) {
487 if(node2->group_id >= 0) {
488 node1->group_id = node2->group_id;
490 node1->group_id = nb_groups;
495 if(node2->group_id < 0) {
496 node2->group_id = node1->group_id;
506 fprintf(stderr, "\n");
510 print_result(list1, list2);
511 file_list_delete(list1);
512 file_list_delete(list2);
514 print_result(list1, 0);
515 file_list_delete(list1);
522 void usage(FILE *out) {
523 fprintf(out, "Usage: finddup [OPTION]... [DIR1 [[and:|not:]DIR2]]\n");
524 fprintf(out, "Version %s (%s)\n", VERSION_NUMBER, UNAME);
525 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");
527 /* 01234567890123456789012345678901234567890123456789012345678901234567890123456789*/
528 fprintf(out, " -h, --help show this help\n");
529 fprintf(out, " -d, --ignore-dots ignore dot files and directories\n");
530 fprintf(out, " -0, --ignore-empty ignore empty files\n");
531 fprintf(out, " -c, --hide-matchings do not show which files in DIR2 corresponds to\n");
532 fprintf(out, " those in DIR1\n");
533 fprintf(out, " -g, --no-group-ids do not show the file groups\n");
534 fprintf(out, " -t, --time-sort sort according to modification time in each group\n");
535 fprintf(out, " -p, --show-progress show progress\n");
536 fprintf(out, " -r, --real-paths show the real file paths\n");
537 fprintf(out, " -i, --same-inodes-are-different\n");
538 fprintf(out, " consider files with same inode as different\n");
540 fprintf(out, "Report bugs and comments to <francois@fleuret.org>.\n");
543 /**********************************************************************/
545 static struct option long_options[] = {
546 { "help", no_argument, 0, 'h' },
547 { "same-inodes-are-different", no_argument, 0, 'i' },
548 { "real-paths", no_argument, 0, 'r' },
549 { "hide-matchings", no_argument, 0, 'c' },
550 { "no-group-ids", no_argument, 0, 'g' },
551 { "time-sort", no_argument, 0, 't' },
552 { "ignore-dots", no_argument, 0, 'd' },
553 { "ignore-empty", no_argument, 0, '0' },
554 { "show-progress", no_argument, 0, 'p' },
558 int main(int argc, char **argv) {
561 setlocale (LC_ALL, "");
563 while ((c = getopt_long(argc, argv, "hircgtd0pm",
564 long_options, NULL)) != -1) {
578 ignore_empty_files = 1;
586 same_inodes_are_different = 1;
611 if(!isatty(STDERR_FILENO)) {
615 if(optind + 2 == argc) {
616 start(argv[optind], argv[optind + 1]);
617 } else if(optind + 1 == argc) {
618 same_inodes_are_different = 1;
619 start(argv[optind], 0);
620 } else if(optind == argc) {
621 same_inodes_are_different = 1;