* to several directories, or files present in one directory and not
* in another one.
*
- * Copyright (c) 2010 Francois Fleuret
+ * Copyright (c) 2010, 2011 Francois Fleuret
* Written by Francois Fleuret <francois@fleuret.org>
*
* This file is part of finddup.
*
*/
-#define VERSION_NUMBER "0.8"
+#define VERSION_NUMBER "1.2.1"
-#define _BSD_SOURCE
+#define _DEFAULT_SOURCE
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/param.h>
+#include <sys/wait.h>
#include <dirent.h>
#include <stdlib.h>
#include <stdio.h>
/* 1M really helps compared to 64k */
#define READ_BUFFER_SIZE (1024 * 1024)
+#define PROGRESS_BUFFER_SIZE 1024
+
typedef int64_t size_sum_t;
/* Yeah, global variables! */
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 show_realpaths = 0; /* 1 show the canonical absolute pathname for
+ printed files */
int show_progress = 0; /* 1 means show a progress bar when we are in a
tty */
int show_groups = 1; /* 1 means to show the group IDs when printing
file names */
-int ignore_same_inode = 0; /* 1 means that comparison between two file
- with same inode will always be false */
+int same_inodes_are_different = 0; /* 1 means that comparison between
+ two files with same inode will
+ always be false */
+
+int sort_by_time = 0; /* 1 means to sort files in each group according
+ to the modification time */
+
+int trim_first = 0; /* remove the first entry in each group */
-int tty_width = -1; /* Positive value means what width to use to show
- the progress bar */
+char *command_to_exec = 0; /* the name of the command to exec for each
+ group of identical files */
+
+char *result_file_prefix = 0; /* The prefix to use to write result */
/********************************************************************/
void *safe_malloc(size_t n) {
void *p = malloc(n);
if (!p && n != 0) {
- printf("Can not allocate memory: %s\n", strerror(errno));
+ fprintf(stderr,
+ "finddup: Can not allocate memory: %s\n", strerror(errno));
exit(EXIT_FAILURE);
}
return p;
struct file_node *next;
char *name;
size_t size;
+ time_t atime, mtime, ctime;
ino_t inode;
int group_id; /* one per identical file content */
int dir_id; /* 1 for DIR1, and 2 for DIR2 */
return 1;
} else {
if(memcmp(buffer1, buffer2, s1)) {
+ /* printf("size_to_read = %d\n", size_to_read); */
close(fd1);
close(fd2);
return 0;
}
} else {
fprintf(stderr,
- "Different read size without error on files of same size.\n");
+ "finddup: Different read size without error on files of same size.\n");
exit(EXIT_FAILURE);
}
}
if(fd1 < 0) {
fprintf(stderr,
- "Can not open \"%s\" error: %s\n",
+ "finddup: Can not open \"%s\" error: %s\n",
f1->name,
strerror(errno));
}
if(fd2 < 0) {
fprintf(stderr,
- "Can not open \"%s\" error: %s\n",
+ "finddup: Can not open \"%s\" error: %s\n",
f2->name,
strerror(errno));
}
int same_files(struct file_node *f1, struct file_node *f2,
char *buffer1, char *buffer2) {
- if(ignore_same_inode && f1->inode == f2->inode) {
+ 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) {
+struct file_node *scan_directory_rec(struct file_node *tail, const char *name) {
DIR *dir;
struct dirent *dir_e;
struct stat sb;
char subname[PATH_MAX + 1];
if(lstat(name, &sb) != 0) {
- fprintf(stderr, "Can not stat \"%s\": %s\n", name, strerror(errno));
+ fprintf(stderr, "finddup: Can not stat \"%s\": %s\n", name, strerror(errno));
exit(EXIT_FAILURE);
}
while((dir_e = readdir(dir))) {
if(!ignore_entry(dir_e->d_name)) {
snprintf(subname, PATH_MAX, "%s/%s", name, dir_e->d_name);
- tail = scan_directory(tail, subname);
+ tail = scan_directory_rec(tail, subname);
}
}
closedir(dir);
tmp->next = tail;
tmp->name = strdup(name);
tmp->size = sb.st_size;
+ tmp->atime = sb.st_atime;
+ tmp->mtime = sb.st_mtime;
+ tmp->ctime = sb.st_ctime;
tmp->inode = sb.st_ino;
tmp->group_id = -1;
tmp->dir_id = -1;
return tail;
}
-void print_file(struct file_node *node) {
+struct file_node *scan_directory(struct file_node *tail, const char *name) {
+ struct file_node *result;
+ int length;
+
+ if(show_progress) {
+ fprintf(stderr, "Scanning '%s' ... ", name);
+ }
+
+ result = scan_directory_rec(tail, name);
+ length = file_list_length(result);
+
+ if(show_progress) {
+ fprintf(stderr, "done (%d file%s).\n",
+ length, (length > 1 ? "s" : ""));
+ }
+
+
+ return result;
+}
+
+void write_one_entry_to_file(FILE *out, 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);
+ fprintf(out, "%d %s\n", node->group_id, tmp);
} else {
- printf("%s\n", tmp);
+ fprintf(out, "%s\n", tmp);
}
} else {
- printf("Can not get the realpath of \"%s\": %s\n",
- node->name,
- strerror(errno));
+ fprintf(stderr,
+ "finddup: 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);
+ fprintf(out, "%d %s\n", node->group_id, node->name);
} else {
- printf("%s\n", node->name);
+ fprintf(out, "%s\n", node->name);
}
}
}
} 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;
+ if(sort_by_time) {
+ if((*f1)->mtime < (*f2)->mtime) {
+ return -1;
+ } else if((*f1)->mtime > (*f2)->mtime) {
+ return 1;
+ } else {
+ return 0;
+ }
} else {
- return 0;
+ if((*f1)->dir_id < (*f2)->dir_id) {
+ return -1;
+ } else if((*f1)->dir_id > (*f2)->dir_id) {
+ return 1;
+ } else {
+ return 0;
+ }
+ }
+ }
+}
+
+void exec_command(int nb, struct file_node **nodes) {
+ char **args;
+ int max_group_size = 0, group_size = 0, m, n, status;
+ pid_t pid;
+
+ for(n = 0; n < nb; n++) {
+ if(n > 0 && nodes[n]->group_id != nodes[n-1]->group_id) {
+ group_size = 0;
+ }
+ group_size++;
+ if(group_size > max_group_size) {
+ max_group_size = group_size;
+ }
+ }
+
+ args = safe_malloc((max_group_size + 2) * sizeof(char *));
+ args[0] = command_to_exec;
+
+ n = 0;
+ while(n < nb) {
+ m = n;
+ if(trim_first) { m++; }
+ while(n < nb && nodes[n]->group_id == nodes[m]->group_id) {
+ if(n >= m) {
+ args[n - m + 1] = nodes[n]->name;
+ }
+ n++;
+ }
+ args[n - m + 1] = 0;
+ pid = fork();
+ if(pid < 0) {
+ } else if(pid == 0) {
+ if(execvp(command_to_exec, args) < 0) {
+ perror("execvp");
+ exit(EXIT_FAILURE);
+ }
+ } else {
+ while(wait(&status) != pid);
+ if(status > 0) {
+ exit(EXIT_FAILURE);
+ }
}
}
+
+ free(args);
}
+void write_groups_in_files(int nb, struct file_node **nodes) {
+ FILE *file = 0;
+ int current_group = -1, n, first_of_group;
+ char filename[PATH_MAX + 1];
+
+ for(n = 0; n < nb; n++) {
+ first_of_group = (n == 0);
+ if(nodes[n]->group_id != current_group) {
+ if(file) { fclose(file); }
+ sprintf(filename, "%s%06d", result_file_prefix, nodes[n]->group_id);
+ file = fopen(filename, "w");
+ current_group = nodes[n]->group_id;
+ printf("Writing %s.\n" , filename);
+ first_of_group = 1;
+ }
+
+ if(!trim_first || !first_of_group) {
+ write_one_entry_to_file(file, nodes[n]);
+ }
+ }
+
+ if(file) { fclose(file); }
+}
void print_result(struct file_node *list1, struct file_node *list2) {
struct file_node *node1, *node2;
struct file_node **nodes;
- int nb, n;
+ int nb, n, first_of_group;
nb = 0;
for(node1 = list1; node1; node1 = node1->next) {
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");
+ if(command_to_exec) {
+ exec_command(nb, nodes);
+ } else if(result_file_prefix) {
+ write_groups_in_files(nb, nodes);
+ } else {
+ for(n = 0; n < nb; n++) {
+ first_of_group = (n == 0);
+ if(n > 0 && nodes[n]->group_id != nodes[n-1]->group_id) {
+ if(!show_groups) {
+ printf("\n");
+ }
+ first_of_group = 1;
+ }
+ if(!trim_first || !first_of_group) {
+ write_one_entry_to_file(stdout, nodes[n]);
+ }
}
- print_file(nodes[n]);
}
free(nodes);
}
-void print_progress(int max, int n, int *pp) {
- int p, k;
- int width;
- if(show_progress && tty_width > 0) {
- 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, "-");
+struct progress_state {
+ int bar_width;
+ int nb_values, value;
+ int last_position;
+};
+
+void print_progress(struct progress_state *state) {
+ int position, k, normalizer;
+ struct winsize win;
+ char buffer[PROGRESS_BUFFER_SIZE];
+ char *s;
+
+ normalizer = (state->nb_values > 1 ? state->nb_values - 1 : 1);
+
+ if(show_progress) {
+ /* We use the previous bar_width to compute the position, so that
+ we avoid doing too many ioctls */
+ position = (state->bar_width * state->value) / normalizer;
+ if(state->bar_width <= 0 || position != state->last_position) {
+ if(!ioctl (STDERR_FILENO, TIOCGWINSZ, (char *) &win)) {
+ /* Something weird is going on if the previous test is wrong */
+ if(win.ws_col >= PROGRESS_BUFFER_SIZE - 3) {
+ state->bar_width = PROGRESS_BUFFER_SIZE - 10;
+ } else {
+ state->bar_width = win.ws_col - 7;
+ }
+ position = (state->bar_width * state->value) / normalizer;
+ state->last_position = position;
+ s = buffer;
+ for(k = 0; k < position; k++) {
+ *(s++) = '+';
+ }
+ for(; k < state->bar_width; k++) {
+ *(s++) = '-';
+ }
+
+ /* We need four % because of the fprintf that follows */
+ sprintf(s, " [%3d%%]\r",
+ (100 * state->value) / normalizer);
+
+ fprintf(stderr, "%s", buffer);
}
- *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;
+ struct progress_state progress_state;
int not_in, found;
int nb_groups, nb_nodes;
- int list1_length, previous_progress;
- struct winsize win;
+ int list1_length;
char *buffer1 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
char *buffer2 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
not_in = 0;
- if(show_progress) {
- if(isatty(STDOUT_FILENO) &&
- !ioctl (STDOUT_FILENO, TIOCGWINSZ, (char *) &win)) {
- tty_width = win.ws_col;
- }
- fprintf(stderr, "Scanning %s ... ", dirname1);
- }
-
list1 = scan_directory(0, dirname1);
+ list1_length = file_list_length(list1);
if(dirname2) {
if(strncmp(dirname2, "not:", 4) == 0) {
} else if(strncmp(dirname2, "and:", 4) == 0) {
dirname2 += 4;
}
- if(show_progress) {
- fprintf(stderr, "%s ... ", dirname2);
- }
list2 = scan_directory(0, dirname2);
} else {
list2 = list1;
}
if(show_progress) {
- fprintf(stderr, "done.\n");
+ fprintf(stderr,
+ "Now looking for identical files (this may take a while).\n");
}
nb_groups = 0;
- previous_progress = -1;
nb_nodes = 0;
- list1_length = file_list_length(list1);
+
+ progress_state.bar_width = -1;
+ progress_state.last_position = -1;
+ progress_state.nb_values = list1_length;
if(not_in) {
for(node1 = list1; node1; node1 = node1->next) {
- print_progress(list1_length, nb_nodes, &previous_progress);
+ progress_state.value = nb_nodes;
+ print_progress(&progress_state);
nb_nodes++;
found = 0;
} else {
for(node1 = list1; node1; node1 = node1->next) {
- print_progress(list1_length, nb_nodes, &previous_progress);
+ progress_state.value = nb_nodes;
+ print_progress(&progress_state);
nb_nodes++;
for(node2 = list2; node2; node2 = node2->next) {
free(buffer2);
}
-void print_help(FILE *out) {
- fprintf(out, "Usage: finddup [OPTION]... DIR1 [[and:|not:]DIR2]\n");
+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, "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");
fprintf(out, "\n");
/* 01234567890123456789012345678901234567890123456789012345678901234567890123456789*/
+ fprintf(out, " -v, --version prints the version number and exit\n");
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, " -c, --hide-matchings do not show which files in DIR2 correspond to\n");
fprintf(out, " those in DIR1\n");
fprintf(out, " -g, --no-group-ids do not show the file groups\n");
+ fprintf(out, " -t, --time-sort sort according to modification time in each group\n");
+ fprintf(out, " -q, --trim-first do not show the first file in each group\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");
+ fprintf(out, " -e <command>, --exec <command>\n");
+ fprintf(out, " execute the provided command for each group of\n");
+ fprintf(out, " identical files, with their names as arguments\n");
+ fprintf(out, " -f <string>, --result-prefix <string>\n");
+ fprintf(out, " for each group of identical files, write one\n");
+ fprintf(out, " result file whose name is the given prefix string\n");
+ fprintf(out, " followed by the group number, and containing\n");
+ fprintf(out, " one filename per line\n");
fprintf(out, "\n");
fprintf(out, "Report bugs and comments to <francois@fleuret.org>.\n");
}
/**********************************************************************/
static struct option long_options[] = {
+ { "version", no_argument, 0, 'v' },
{ "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' },
+ { "time-sort", no_argument, 0, 't' },
+ { "trim-first", no_argument, 0, 'q' },
{ "ignore-dots", no_argument, 0, 'd' },
{ "ignore-empty", no_argument, 0, '0' },
{ "show-progress", no_argument, 0, 'p' },
+ { "exec", 1, 0, 'e' },
+ { "result-prefix", 1, 0, 'f' },
{ 0, 0, 0, 0 }
};
setlocale (LC_ALL, "");
- while ((c = getopt_long(argc, argv, "hircgd0p",
+ while ((c = getopt_long(argc, argv, "vhircgtqd0pme:f:",
long_options, NULL)) != -1) {
switch (c) {
+ case 'v':
+ printf("finddup version %s (%s)\n", VERSION_NUMBER, UNAME);
+ exit(EXIT_SUCCESS);
+ break;
+
case 'h':
- print_help(stdout);
+ usage(stdout);
exit(EXIT_SUCCESS);
break;
break;
case 'i':
- ignore_same_inode = 1;
+ same_inodes_are_different = 1;
break;
case 'g':
show_groups = 0;
break;
+ case 't':
+ sort_by_time = 1;
+ break;
+
+ case 'q':
+ trim_first = 1;
+ break;
+
case 'p':
show_progress = 1;
break;
show_hits = 0;
break;
+ case 'e':
+ if(command_to_exec != 0) {
+ free(command_to_exec);
+ }
+ command_to_exec = strdup(optarg);
+ break;
+
+ case 'f':
+ if(result_file_prefix != 0) {
+ free(result_file_prefix);
+ }
+ result_file_prefix = strdup(optarg);
+ show_groups = 0;
+ break;
+
default:
+ usage(stderr);
exit(EXIT_FAILURE);
}
}
+ if(!isatty(STDERR_FILENO)) {
+ show_progress = 0;
+ }
+
if(optind + 2 == argc) {
start(argv[optind], argv[optind + 1]);
} else if(optind + 1 == argc) {
- ignore_same_inode = 1;
+ same_inodes_are_different = 1;
start(argv[optind], 0);
+ } else if(optind == argc) {
+ same_inodes_are_different = 1;
+ start(".", 0);
} else {
- print_help(stderr);
+ usage(stderr);
exit(EXIT_FAILURE);
}
projects / finddup.git / blobdiff