JRF, you're right. Thanks. We indeed should use
defined getpwuid
But then I would use positive logic in the ternary
# perl -MFile::Find -e'find(sub{-f and push@{$f{-s$_}},[$File::Find::name,(stat$_)[4]]},@ARGV);for$s(sort{$b<=>$a}keys%f){printf"%4s %9d %s\n",defined getpwuid$_->[1]?"":"ORPH",$s,$_->[0] for sort@{$f{$s}}}' /home11
A small breadown:
> # perl
the command invokation
> -MFile::Find
use modules File::Find. This is equivalent to
use File::Find:
in a script
> -e
command line option.
-e will use the next arg as a chunk of perl code
> find (
find is a function from the File::Find module, and has several invocation methods, basically that sums down to
find (function, directories)
as function can be a (reference to a) named sub or an anonymous sub, I always prefer to inline it in the call using sub { }
The sub defines the actions being taken on all entries found
> sub {
> -f and push @{$f{-s$_}},[$File::Find::name,(stat$_)[4]]
-f is a file test operator, defaulting to $_. If you want to do that explicitely, add it:
# only scan *files*, not special files, symlinks, directories and fifos
-f $_ or return;
# Note that you have to use 'return', and not 'next' as you are leaving a sub, not skipping to the next file in a for or while loop. A common pitfall for find ()
# Now push the file info in a hash of lists
# The hash is `indexed' by the filesize (-s)
# For each entry, I push an anonymous list with two entries:
# 1. The full file name ($File::Find::name)
# 2. The owner of the file (stat $_)[4]
# see perldoc -f stat
push @{$f{-s$_}}, [ $File::Find::name, (stat$_)[4] ];
I chose to use the file size as base index for the hash, so descending sorting on the file size is now dead easy
> }, @ARGV);
} is end of sub, @ARGV is the directories. @ARGV is the rest of the command line arguments
So, now I have gathered all the info I need in a hash, I must show it
> for $s (sort {$b <=> $a} keys %f) {
$s is iterating over the numeriacally descending sorted (see perldoc -f sort) keys of the hash. As the keys are the file sizes, this meets your quest
> printf "%4s %9d %s\n", defined getpwuid $_->[1] ? "" : "ORPH" , $s, $_->[0] for sort @{$f{$s}}}
OK, that is a lot on one line. Lets break that down some more to something more extendable
# Get me the (alphabetically sorted) list of files that have this size
my @files = sort @{$f{$s}};
# Now iterate over it
foreach my $entry (@files) {
# Each entry had the name and the uid
# in an anonymous list
my ($file, $uid) = @$_;
if (defined getwpuid $uid) {
# This file has a known user
printf "%9d %s\n", $s, $file;
}
else {
# ALARM: Orphaned file
printf "ORPHANED: %9d %s\n", $s, $file;
}
}
>' /home11
As I used @ARGV for the directories to be found, I need to pass these as extra arguments. I could also have chosen to hardcode them in the find command, but the way it is both easier to move it to a script and to change it with command line editing
Hope that helped.
One more note. Files without a named user from /etc/passwd do not have to be orphaned by means of the user being removed from the system. It can also be the uid of a file being retreived from a cpio or tar (sw depot) file that was installed from another system with uid's that do not match your's
Enjoy, Have FUN! H.Merijn
Enjoy, Have FUN! H.Merijn
