#include <locale.h>
#include <getopt.h>
#include <fcntl.h>
-#ifdef WITH_MD5
-#include <openssl/md5.h>
-#endif
/* 1M really helps compared to 64k */
#define READ_BUFFER_SIZE (1024 * 1024)
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
-
/********************************************************************/
/* malloc with error checking. */
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) {
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(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)) {
+ /* printf("size_to_read = %d\n", size_to_read); */
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,
tmp->inode = sb.st_ino;
tmp->group_id = -1;
tmp->dir_id = -1;
-#ifdef WITH_MD5
- tmp->md5_computed = 0;
-#endif
tail = tmp;
}
}
not_in = 0;
if(show_progress) {
- fprintf(stderr, "Scanning %s ... ", dirname1);
+ fprintf(stderr, "Scanning '%s' ... ", dirname1);
}
list1 = scan_directory(0, dirname1);
list1_length = file_list_length(list1);
+ if(show_progress) {
+ fprintf(stderr, "done (%d file%s).\n",
+ list1_length, (list1_length > 1 ? "s" : ""));
+ }
+
if(dirname2) {
if(strncmp(dirname2, "not:", 4) == 0) {
not_in = 1;
dirname2 += 4;
}
if(show_progress) {
- fprintf(stderr, "%s ... ", dirname2);
+ fprintf(stderr, "Scanning '%s' ... ", dirname2);
}
list2 = scan_directory(0, dirname2);
+ if(show_progress) {
+ list2_length = file_list_length(list2);
+ fprintf(stderr, "done (%d file%s).\n",
+ list2_length, (list2_length > 1 ? "s" : ""));
+ }
} else {
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");
+ fprintf(stderr, "Now looking for identical files (this may take a while).\n");
}
nb_groups = 0;
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");
}
{ "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 }
};
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);