$input_dir =~ s{^\.\./}{/};
[download]

See perlretut.

However, it seems a bit strange to want to do it this way. If you want to do some advanced filename manipulation, I suggest Path::Class or Path::Tiny, or at the very least the core module File::Spec. All three of these also have the added advantage of being portable across OSes (Path::Tiny only does *NIX and Win32, though).

use File::Spec::Functions qw/splitdir updir catdir/;
my @input_dir = splitdir $input_dir;
shift @input_dir if $input_dir[0] eq updir;
$input_dir = catdir(@input_dir);
# $input_dir is now "set/projects/all/files"
[download]

Update: Corrected {^../} to {^\.\./}.

You know monks can be very susceptibles after many years passed in their cells sudiyng Scriptures.

That said your output come from your regular expression: /^LOCATION:\s*(\S+)/ captures ../set/projects/all/files if given LOCATION:../set/projects/all/files as input.

If you do not want the two dots to be captured you can put it outside capturing parens. Dots must be escaped in regexes because they are special chars. So I guess that /^LOCATION:\s*\.\.(\S+)/ shold not capture the two unwanted dots.

L*

How can i remove the .. from the directory line and read only the location of the directory

Do it after?

A more generic parser

my %action = (
    LOCATION => sub {
        my( $dir ) = @_;
        $dir =~ s{^\.\.}{};
        print $dir;
    },
);

... while ...

/^         # start of line
\s*        # optional padding
( [^:]+ ): # $1 key, not ':'
\s*
([^\r\n]*) # $2 value, not line end
[\r\n]*    # end of line
/xm
and do {
    my( $key, $value ) = @_;
    my $action = $parser{ $key } || $parser{default};
    $action->($value);
};
[download]