Hard links are counted as duplicates

[blob79]

Hard linked files shouldn't be counted as duplicates as they do not take up space on disk.

$ echo a > a
$ ln a b
$ cp a c
$ stat a b c  | grep Inode
Device: 10302h/66306d	Inode: 25463856    Links: 2
Device: 10302h/66306d	Inode: 25463856    Links: 2
Device: 10302h/66306d	Inode: 25463858    Links: 1

$ cargo run dup  --directories ~/Development/czkawka/czkawka_cli -m1 -f /dev/stdout | grep -A99 "Files with"
-------------------------------------------------Files with same hashes-------------------------------------------------
Found 2 duplicated files which in 1 groups which takes 4 B.

---- Size 2 B (2) - 3 files
/home/thomas/Development/czkawka/czkawka_cli/a
/home/thomas/Development/czkawka/czkawka_cli/c
/home/thomas/Development/czkawka/czkawka_cli/b

$ rm c
$ cargo run dup  --directories ~/Development/czkawka/czkawka_cli -m1 -f /dev/stdout | grep -A99 "Files with"
-------------------------------------------------Files with same hashes-------------------------------------------------
Found 1 duplicated files which in 1 groups which takes 2 B.

---- Size 2 B (2) - 2 files
/home/thomas/Development/czkawka/czkawka_cli/a
/home/thomas/Development/czkawka/czkawka_cli/b

It also looks like the duplicate disk space counting goes wrong in this case it should be 2. If you double count the hard link it should 3 not 4.

[TANDEXX]

Command ls or any file manager see hard links as normal files. Program can't check do it is hard link, for program it's regular file and this program can't check this too.

[Sbgodin]

@TANDEXX, not exactly. ls command knows and shows when a file was hardlinked – indeed there are several ways to reach the file – by counting the reference number:

$ echo "hello" > file_1

$ ls -l file_1
-rw-r--r-- 1 user group 6 11 févr. 00:03 file_1

$ ln file_1 file_2
$ ls -l file_1 file 2
-rw-r--r-- 2 user group 6 11 févr. 00:03 file_1
-rw-r--r-- 2 user group 6 11 févr. 00:03 file_2

The two files have now a reference count of 2.

find . -samefile file_1
./file_2
./file_1

The find command knows it too.

[blob79]

You can look at the inode with:

❯ ls -i
13644513 a
13644513 b

ls man page:
-i, --inode
print the index number of each file

#234 implements the fix.