Re: Finding Hard links

I don't know it there's a module available to do that. But I suppose that as a general rule it won't be an easy thing since, AFAIK while filesystem and related system calls return relevant information in the direction filename => inode, they do not do the opposite. So you can gather them from comparing files. Of course this is made easier by the fact that

you only have to compare files on the same fs,
you only have to compare files with the same number, larger than 1, of (hard) links as known from stat, which hopefully are not that many.

Update: ok, here I try my hand at this with a minimal example/proof of concept:

#!/usr/bin/perl -ln

use strict;
use warnings;

BEGIN { @ARGV='find . -xdev|' }

our %files;

my ($ino, $nlink)=(stat)[1,3];
next unless $nlink and $nlink > 1;
if ($nlink == push @{ $files{$ino} }, $_) {
    print "$_ => $ino" for @{ $files{$ino} };
    delete $files{$ino};
}

__END__
[download]

Comment on Re: Finding Hard links Select or Download Code

Replies are listed 'Best First'.
Re^2: Finding Hard links by powerman (Friar) on Nov 09, 2005 at 03:54 UTC
Original question was about perl module... but if you start using `find` command, then: my $original_file = '/path/to/file'; my $dir = '/where/to/find/hardlinks/'; chomp for @hard_links = `find \Q$dir\E -inum \$(find \Q$original_file\E -printf "%i")`; [download]	[reply] [d/l]
Re^3: Finding Hard links by blazar (Canon) on Nov 09, 2005 at 14:20 UTC
Oh, well of course! I used an external `find` command because as I stressed, I just wanted to keep minimal and using File::Find, as I would most probably do in a more realistic case¹, was not substantial for the technique I wanted to explore... You'll also notice that I used `-ln` on a slightly longer script than I would usually do in "production", whatever that is. But in the above you're fundamentally using Perl as a shell script. That is really "kinda too much". Incidentally you're using the "inner" `find` just to print the inode number. In that case the external `stat` command would suffice: `$ stat -c %i work/ 18972756` [download] OTOH I didn't know about `-inum`, so I thank you - I know it's out there in the docs, but I tend to learn by example... Also, as a final observation, I think that indeed the OP was really asking about how to find hard links sharing the same inode as a given file, that is what you actually do. But if you want to search a whole directory hierarchy like I did, a "pure-find" approach along the lines of your example would be unpractical, since I think it would necessarily take two nested `find`'s. ¹ And if I called an external find cmd, I'd probably use an explicit open.	[reply] [d/l] [select]