/*
- * finddup is a simple utility to display the files and directories
- * according to their total disk occupancy.
+ * finddup is a simple utility to find duplicated files, files common
+ * to several directories, or files present in one directory and not
+ * in another one.
*
* Copyright (c) 2010 Francois Fleuret
* Written by Francois Fleuret <francois@fleuret.org>
*
*/
+#define VERSION_NUMBER "0.9"
+
#define _BSD_SOURCE
#include <sys/types.h>
#include <sys/stat.h>
+#include <sys/param.h>
#include <dirent.h>
#include <stdlib.h>
#include <stdio.h>
#include <locale.h>
#include <getopt.h>
#include <fcntl.h>
+#ifdef WITH_MD5
+#include <openssl/md5.h>
+#endif
-#define BUFFER_SIZE 4096
+/* 1M really helps compared to 64k */
+#define READ_BUFFER_SIZE (1024 * 1024)
typedef int64_t size_sum_t;
int ignore_dotfiles = 0; /* 1 means ignore files and directories
starting with a dot */
-int forced_width = 0; /* -1 means no width limit, strictly positive
- means limit, 0 means not active */
+int ignore_empty_files = 0; /* 1 means ignore empty files */
+
+int show_realpaths = 0; /* 1 means ignore files and directories
+ starting with a dot */
-int forced_height = 0; /* -1 means no height limit, strictly positive
- means limit, 0 means not active */
+int show_progress = 0; /* 1 means show a progress bar when we are in a
+ tty */
-int fancy_size_display = 0; /* 1 means to use floating values with K, M and G
- as units */
+int show_hits = 1; /* 1 means to show the files from dir2
+ corresponding to the ones from dir1 */
-int reverse_sorting = 0; /* 1 means to show the large ones first */
+int show_groups = 1; /* 1 means to show the group IDs when printing
+ file names */
-int show_top = 0; /* 1 means to show the top of the sorted list
- instead of the bottom */
+int same_inodes_are_different = 0; /* 1 means that comparison between
+ two files with same inode will
+ always be false */
+
+#ifdef WITH_MD5
+int use_md5 = 0; /* 1 means we keep an MD5 signature for each file */
+#endif
/********************************************************************/
return
strcmp(name, ".") == 0 ||
strcmp(name, "..") == 0 ||
- (ignore_dotfiles && name[0] == '.');
+ (ignore_dotfiles && name[0] == '.' && name[1] != '/');
}
-void print_size_sum(size_sum_t s) {
- char tmp[100];
- char *a = tmp + sizeof(tmp)/sizeof(char);
- *(--a) = '\0';
- if(s) {
- while(s) {
- *(--a) = s%10 + '0';
- s /= 10;
+/**********************************************************************/
+
+struct file_node {
+ struct file_node *next;
+ char *name;
+ size_t size;
+ ino_t inode;
+ int group_id; /* one per identical file content */
+ int dir_id; /* 1 for DIR1, and 2 for DIR2 */
+#ifdef WITH_MD5
+ int md5_computed;
+ unsigned char md5[MD5_DIGEST_LENGTH];
+#endif
+};
+
+void file_list_delete(struct file_node *head) {
+ struct file_node *next;
+ while(head) {
+ next = head->next;
+ free(head->name);
+ free(head);
+ head = next;
+ }
+}
+
+int file_list_length(struct file_node *head) {
+ int l = 0;
+ while(head) {
+ l++;
+ head = head->next;
+ }
+ return l;
+}
+
+/**********************************************************************/
+
+int same_content(struct file_node *f1, struct file_node *f2,
+ char *buffer1, char *buffer2) {
+ int fd1, fd2, s1, s2;
+
+#ifdef WITH_MD5
+ MD5_CTX c1, c2;
+
+ if(use_md5) {
+ if(f1->md5_computed && f2->md5_computed) {
+ if(!memcmp(f1->md5, f2->md5, MD5_DIGEST_LENGTH)) {
+ return 0;
+ }
+ } else {
+ if(!f1->md5_computed) {
+ MD5_Init(&c1);
+ }
+ if(!f2->md5_computed) {
+ MD5_Init(&c2);
+ }
+ }
+ }
+#endif
+
+ fd1 = open(f1->name, O_RDONLY);
+ fd2 = open(f2->name, O_RDONLY);
+
+ if(fd1 >= 0 && fd2 >= 0) {
+ while(1) {
+ s1 = read(fd1, buffer1, READ_BUFFER_SIZE);
+ s2 = read(fd2, buffer2, READ_BUFFER_SIZE);
+
+ if(s1 < 0 || s2 < 0) {
+ close(fd1);
+ close(fd2);
+ return 0;
+ }
+
+ if(s1 == s2) {
+ if(s1 == 0) {
+ close(fd1);
+ close(fd2);
+#ifdef WITH_MD5
+ if(use_md5) {
+ if(!f1->md5_computed) {
+ MD5_Final(f1->md5, &c1);
+ f1->md5_computed = 1;
+ }
+ if(!f2->md5_computed) {
+ MD5_Final(f2->md5, &c2);
+ f2->md5_computed = 1;
+ }
+ }
+#endif
+ return 1;
+ } else {
+ if(memcmp(buffer1, buffer2, s1)) {
+ close(fd1);
+ close(fd2);
+ return 0;
+ }
+#ifdef WITH_MD5
+ if(use_md5) {
+ if(!f1->md5_computed) {
+ MD5_Update(&c1, buffer1, s1);
+ }
+ if(!f2->md5_computed) {
+ MD5_Update(&c2, buffer2, s2);
+ }
+ }
+#endif
+ }
+ } else {
+ fprintf(stderr,
+ "Different read size without error on files of same size.\n");
+ exit(EXIT_FAILURE);
+ }
}
} else {
- *(--a) = '0';
+
+ if(fd1 < 0) {
+ fprintf(stderr,
+ "Can not open \"%s\" error: %s\n",
+ f1->name,
+ strerror(errno));
+ }
+
+ if(fd2 < 0) {
+ fprintf(stderr,
+ "Can not open \"%s\" error: %s\n",
+ f2->name,
+ strerror(errno));
+ }
+
+ exit(EXIT_FAILURE);
}
- printf(a);
}
-size_sum_t file_or_dir_size(const char *name) {
+int same_files(struct file_node *f1, struct file_node *f2,
+ char *buffer1, char *buffer2) {
+ if(same_inodes_are_different && f1->inode == f2->inode) {
+ return 0;
+ }
+
+ return f1->size == f2->size && same_content(f1, f2, buffer1, buffer2);
+}
+
+/**********************************************************************/
+
+struct file_node *scan_directory(struct file_node *tail, const char *name) {
DIR *dir;
struct dirent *dir_e;
- struct stat dummy;
- size_sum_t result;
- char subname[BUFFER_SIZE];
-
- result = 0;
+ struct stat sb;
+ struct file_node *tmp;
+ char subname[PATH_MAX + 1];
- if(lstat(name, &dummy) != 0) {
- printf("Can not stat %s: %s\n", name, strerror(errno));
+ if(lstat(name, &sb) != 0) {
+ fprintf(stderr, "Can not stat \"%s\": %s\n", name, strerror(errno));
exit(EXIT_FAILURE);
}
- if(S_ISLNK(dummy.st_mode)) {
- return 0;
+ if(S_ISLNK(sb.st_mode)) {
+ return tail;
}
dir = opendir(name);
if(dir) {
while((dir_e = readdir(dir))) {
if(!ignore_entry(dir_e->d_name)) {
- snprintf(subname, BUFFER_SIZE, "%s/%s", name, dir_e->d_name);
- result += file_or_dir_size(subname);
+ snprintf(subname, PATH_MAX, "%s/%s", name, dir_e->d_name);
+ tail = scan_directory(tail, subname);
}
}
closedir(dir);
} else {
- if(S_ISREG(dummy.st_mode)) {
- result += dummy.st_size;
+ if(S_ISREG(sb.st_mode)) {
+ if(!ignore_entry(name)) {
+ if(!ignore_empty_files || sb.st_size > 0) {
+ tmp = safe_malloc(sizeof(struct file_node));
+ tmp->next = tail;
+ tmp->name = strdup(name);
+ tmp->size = sb.st_size;
+ tmp->inode = sb.st_ino;
+ tmp->group_id = -1;
+ tmp->dir_id = -1;
+#ifdef WITH_MD5
+ tmp->md5_computed = 0;
+#endif
+ tail = tmp;
+ }
+ }
}
}
- return result;
+ return tail;
}
-/**********************************************************************/
-
-struct file_with_size {
- char *filename;
- size_t size;
- struct file_with_size *next;
- int error;
-};
+void print_file(struct file_node *node) {
+ char tmp[PATH_MAX + 1];
+ if(show_realpaths) {
+ if(realpath(node->name, tmp)) {
+ if(show_groups) {
+ printf("%d %s\n", node->group_id, tmp);
+ } else {
+ printf("%s\n", tmp);
+ }
+ } else {
+ printf("Can not get the realpath of \"%s\": %s\n",
+ node->name,
+ strerror(errno));
+ exit(EXIT_FAILURE);
+ }
+ } else {
+ if(show_groups) {
+ printf("%d %s\n", node->group_id, node->name);
+ } else {
+ printf("%s\n", node->name);
+ }
+ }
+}
-struct file_list {
- struct file_with_size *head, *tail;
-};
+int compare_nodes(const void *x1, const void *x2) {
+ const struct file_node **f1, **f2;
-void create(struct file_list *list, const char *name) {
- list->head = safe_malloc(sizeof(struct file_with_size));
- list->tail = 0;
- list->head->filename = strdup(name);
- list->head->size = file_or_dir_size(name);
- list->head->next = 0;
-}
+ f1 = (const struct file_node **) x1;
+ f2 = (const struct file_node **) x2;
-void destroy(struct file_with_size *node) {
- struct file_with_size *next;
- while(node) {
- next = node->next;
- free(node->filename);
- free(node);
- node = next;
+ if((*f1)->group_id < (*f2)->group_id) {
+ return -1;
+ } else if((*f1)->group_id > (*f2)->group_id) {
+ return 1;
+ } else {
+ if((*f1)->dir_id < (*f2)->dir_id) {
+ return -1;
+ } else if((*f1)->dir_id > (*f2)->dir_id) {
+ return 1;
+ } else {
+ return 0;
+ }
}
}
-/**********************************************************************/
-int same_size(struct file_with_size *f1, struct file_with_size *f2) {
- return f1->size == f2->size;
-}
+void print_result(struct file_node *list1, struct file_node *list2) {
+ struct file_node *node1, *node2;
+ struct file_node **nodes;
+ int nb, n;
-int same_content(struct file_with_size *f1, struct file_with_size *f2) {
- int fd1, fd2, eq, s1, s2;
- char buffer1[BUFFER_SIZE], buffer2[BUFFER_SIZE];
+ nb = 0;
+ for(node1 = list1; node1; node1 = node1->next) {
+ if(node1->group_id >= 0) { nb++; }
+ }
- fd1 = open(f1->filename, O_RDONLY);
- fd2 = open(f2->filename, O_RDONLY);
- eq = 1;
+ if(list2) {
+ for(node2 = list2; node2; node2 = node2->next) {
+ if(node2->group_id >= 0) { nb++; }
+ }
+ }
- if(fd1 >= 0 && fd2 >= 0) {
- while(eq) {
- s1 = read(fd1, buffer1, BUFFER_SIZE);
- if(s1 < 0) {
- f1->error = 1;
- eq = 0;
- }
- s2 = read(fd2, buffer2, BUFFER_SIZE);
- if(s2 < 0) {
- f2->error = 1;
- eq = 0;
- }
- if(s1 > 0 && s2 > 0) {
- if(s1 == s2) {
- eq = (strncmp(buffer1, buffer2, s1) == 0);
- } else {
- fprintf(stderr,
- "Different read size without error on files of same size.\n");
- exit(EXIT_FAILURE);
- }
- }
+ nodes = safe_malloc(nb * sizeof(struct file_node *));
+
+ n = 0;
+ for(node1 = list1; node1; node1 = node1->next) {
+ if(node1->group_id >= 0) {
+ nodes[n++] = node1;
}
}
- if(fd1 >= 0) {
- close(fd1);
- } else {
- f1->error = 1;
- eq = 0;
+ if(list2) {
+ for(node2 = list2; node2; node2 = node2->next) {
+ if(node2->group_id >= 0) {
+ nodes[n++] = node2;
+ }
+ }
}
- if(fd2 >= 0) {
- close(fd2);
- } else {
- f2->error = 1;
- eq = 0;
+ qsort(nodes, nb, sizeof(struct file_node *), compare_nodes);
+
+ for(n = 0; n < nb; n++) {
+ if(!show_groups && n > 0 && nodes[n]->group_id != nodes[n-1]->group_id) {
+ printf("\n");
+ }
+ print_file(nodes[n]);
}
- return eq;
+ free(nodes);
}
-int same_files(struct file_with_size *f1, struct file_with_size *f2) {
- return same_size(f1, f2) && same_content(f1, f2);
+void print_progress(int max, int n, int *pp) {
+ int p, k;
+ int width, tty_width;
+ struct winsize win;
+
+ if(show_progress &&
+ isatty(STDOUT_FILENO) &&
+ !ioctl (STDOUT_FILENO, TIOCGWINSZ, (char *) &win)) {
+ tty_width = win.ws_col;
+ width = tty_width - 7;
+ p = (width * n) / (max - 1);
+ if(p > *pp) {
+ for(k = 0; k < p; k++) {
+ fprintf(stderr, "+");
+ }
+ for(; k < width; k++) {
+ fprintf(stderr, "-");
+ }
+ *pp = p;
+ p = (100 * n) / (max - 1);
+ fprintf(stderr, " [%3d%%]\r", p);
+ }
+ }
}
-/**********************************************************************/
+void start(const char *dirname1, const char *dirname2) {
+ struct file_node *list1, *list2;
+ struct file_node *node1, *node2;
+ int not_in, found;
+ int nb_groups, nb_nodes;
+ int list1_length, list2_length, previous_progress;
-struct file_with_size *scan_directory(struct file_with_size *tail, const char *name) {
- DIR *dir;
- struct dirent *dir_e;
- struct stat dummy;
- struct file_with_size *tmp;
- char subname[BUFFER_SIZE];
+ char *buffer1 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
+ char *buffer2 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
- if(lstat(name, &dummy) != 0) {
- fprintf(stderr, "Can not stat %s: %s\n", name, strerror(errno));
- exit(EXIT_FAILURE);
- }
+ not_in = 0;
- if(S_ISLNK(dummy.st_mode)) {
- return tail;
+ if(show_progress) {
+ fprintf(stderr, "Scanning %s ... ", dirname1);
}
- dir = opendir(name);
+ list1 = scan_directory(0, dirname1);
- if(dir) {
- while((dir_e = readdir(dir))) {
- if(!ignore_entry(dir_e->d_name)) {
- snprintf(subname, BUFFER_SIZE, "%s/%s", name, dir_e->d_name);
- tail = scan_directory(tail, subname);
- }
+ list1_length = file_list_length(list1);
+
+ if(dirname2) {
+ if(strncmp(dirname2, "not:", 4) == 0) {
+ not_in = 1;
+ /* groups are not computed in the not: mode */
+ show_groups = 0;
+ dirname2 += 4;
+ } else if(strncmp(dirname2, "and:", 4) == 0) {
+ dirname2 += 4;
}
- closedir(dir);
+ if(show_progress) {
+ fprintf(stderr, "%s ... ", dirname2);
+ }
+ list2 = scan_directory(0, dirname2);
} else {
- if(S_ISREG(dummy.st_mode)) {
- tmp = safe_malloc(sizeof(struct file_with_size));
- tmp->next = tail;
- tmp->filename = strdup(name);
- tmp->size = file_or_dir_size(name);
- tail = tmp;
+ list2 = list1;
+ }
+
+ if(show_progress) {
+ fprintf(stderr, "done.\n");
+ fprintf(stderr,
+ "%s: %d file%s.\n",
+ dirname1, list1_length, (list1_length > 1 ? "s" : ""));
+ if(dirname2) {
+ list2_length = file_list_length(list2);
+ fprintf(stderr,
+ "%s: %d file%s.\n",
+ dirname2, list2_length, (list2_length > 1 ? "s" : ""));
}
+ fprintf(stderr, "Now looking for identical files.\n");
}
- return tail;
-}
+ nb_groups = 0;
+ previous_progress = -1;
+ nb_nodes = 0;
-void start(const char *dirname1, const char *dirname2) {
- struct file_with_size *list1, *list2;
- struct file_with_size *node;
- list1 = scan_directory(0, dirname1);
- list2 = scan_directory(0, dirname2);
+ if(not_in) {
+ for(node1 = list1; node1; node1 = node1->next) {
+ print_progress(list1_length, nb_nodes, &previous_progress);
+ nb_nodes++;
+
+ found = 0;
+
+ for(node2 = list2; !found && node2; node2 = node2->next) {
+ if(same_files(node1, node2, buffer1, buffer2)) {
+ found = 1;
+ }
+ }
+
+ if(!found) {
+ if(show_realpaths) {
+ printf("%s\n", realpath(node1->name, 0));
+ } else {
+ printf("%s\n", node1->name);
+ }
+ }
+ }
- for(node = list1; node; node = node->next) {
- printf("list1: %s\n", node->filename);
+ } else {
+ for(node1 = list1; node1; node1 = node1->next) {
+ print_progress(list1_length, nb_nodes, &previous_progress);
+ nb_nodes++;
+
+ for(node2 = list2; node2; node2 = node2->next) {
+ if(node1->group_id < 0 || node2->group_id < 0) {
+ if(same_files(node1, node2, buffer1, buffer2)) {
+ if(node1->group_id < 0) {
+ if(node2->group_id >= 0) {
+ node1->group_id = node2->group_id;
+ } else {
+ node1->group_id = nb_groups;
+ node1->dir_id = 1;
+ nb_groups++;
+ }
+ }
+ if(node2->group_id < 0) {
+ node2->group_id = node1->group_id;
+ node2->dir_id = 2;
+ }
+ }
+ }
+ }
+ }
}
- for(node = list2; node; node = node->next) {
- printf("list2: %s\n", node->filename);
+ if(show_progress) {
+ fprintf(stderr, "\n");
}
- destroy(list1);
- destroy(list2);
+ if(dirname2) {
+ print_result(list1, list2);
+ file_list_delete(list1);
+ file_list_delete(list2);
+ } else {
+ print_result(list1, 0);
+ file_list_delete(list1);
+ }
+
+ free(buffer1);
+ free(buffer2);
+}
+
+void usage(FILE *out) {
+ fprintf(out, "Usage: finddup [OPTION]... DIR1 [[and:|not:]DIR2]\n");
+ fprintf(out, "Version %s (%s)\n", VERSION_NUMBER, UNAME);
+ 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");
+ fprintf(out, "\n");
+ /* 01234567890123456789012345678901234567890123456789012345678901234567890123456789*/
+ fprintf(out, " -h, --help show this help\n");
+ fprintf(out, " -d, --ignore-dots ignore dot files and directories\n");
+ fprintf(out, " -0, --ignore-empty ignore empty files\n");
+ fprintf(out, " -c, --hide-matchings do not show which files in DIR2 corresponds to\n");
+ fprintf(out, " those in DIR1\n");
+ fprintf(out, " -g, --no-group-ids do not show the file groups\n");
+ fprintf(out, " -p, --show-progress show progress\n");
+ fprintf(out, " -r, --real-paths show the real file paths\n");
+ fprintf(out, " -i, --same-inodes-are-different\n");
+ fprintf(out, " consider files with same inode as different\n");
+#ifdef WITH_MD5
+ fprintf(out, " -m, --md5 use MD5 hashing\n");
+#endif
+ fprintf(out, "\n");
+ fprintf(out, "Report bugs and comments to <francois@fleuret.org>.\n");
}
/**********************************************************************/
+static struct option long_options[] = {
+ { "help", no_argument, 0, 'h' },
+ { "same-inodes-are-different", no_argument, 0, 'i' },
+ { "real-paths", no_argument, 0, 'r' },
+ { "hide-matchings", no_argument, 0, 'c' },
+ { "no-group-ids", no_argument, 0, 'g' },
+ { "ignore-dots", no_argument, 0, 'd' },
+ { "ignore-empty", no_argument, 0, '0' },
+ { "show-progress", no_argument, 0, 'p' },
+ { "md5", no_argument, 0, 'm' },
+ { 0, 0, 0, 0 }
+};
+
int main(int argc, char **argv) {
int c;
- struct file_with_size *root;
-
- root = 0;
setlocale (LC_ALL, "");
- while (1) {
- c = getopt(argc, argv, "h");
- if (c == -1)
- break;
-
+ while ((c = getopt_long(argc, argv, "hircgd0pm",
+ long_options, NULL)) != -1) {
switch (c) {
case 'h':
- printf("Usage: finddup [OPTION]... [FILE]...\n");
- printf("Report bugs and comments to <francois@fleuret.org>\n");
+ usage(stdout);
exit(EXIT_SUCCESS);
break;
+ case 'd':
+ ignore_dotfiles = 1;
+ break;
+
+ case '0':
+ ignore_empty_files = 1;
+ break;
+
+ case 'r':
+ show_realpaths = 1;
+ break;
+
+ case 'i':
+ same_inodes_are_different = 1;
+ break;
+
+ case 'g':
+ show_groups = 0;
+ break;
+
+ case 'p':
+ show_progress = 1;
+ break;
+
+ case 'c':
+ show_hits = 0;
+ break;
+
+ case 'm':
+#ifdef WITH_MD5
+ use_md5 = 1;
+#else
+ fprintf(stderr,
+ "finddup has not been compiled with MD5 hashing.\n");
+ usage(stderr);
+ exit(EXIT_FAILURE);
+#endif
+ break;
+
default:
+ usage(stderr);
exit(EXIT_FAILURE);
}
}
- if(optind + 1 < argc) {
+ if(optind + 2 == argc) {
start(argv[optind], argv[optind + 1]);
+ } else if(optind + 1 == argc) {
+ same_inodes_are_different = 1;
+ start(argv[optind], 0);
} else {
- fprintf(stderr, "%s [OPTIONS] <dir1> <dir2>\n", argv[0]);
+ usage(stderr);
exit(EXIT_FAILURE);
}