perl overload file name download - perl

Question

I need to be able to propose files to be downloaded but i have to read and print the file in my CGI. I tried to go for :
#!/usr/bin/perl -w
use strict;
push( #INC, $lib_directory );
require 'lib_utils.pl';
dl_file('/tmp/final.pdf');
as main page (dl.pl) and
sub dl_file {
my ($file) = #_;
if ( ! -e $file) {
print "file does not exist";
return 0;
}
my $content = read_file( $file, binmode => ':utf8' ) ;
$file =~ m#(.*)([^/]*)$#;
my $directory = $1;
my $filename = $2;
chdir $directory;
my $form = new CGI;
print $form->header(
-type => 'application/octet-stream',
-attachment => $filename,
-filename => $filename,
-Content-Disposition => "attachment; filename=$filename",
);
$form->print($content);
return 1;
}
for the called function. Funny thing is, this code workes just fine if i dont go for a sub and have all the code in dl.pl BUT as soon as i move the code in a sub, the downloaded file is called after the script (ie dl.pl)
How would you change it or how would you do ?
Thanks in advance for your help

Answer 1

Your line
$file =~ m#(.*)([^/]*)$#
will leave $1 containing the whole of $file and $2 empty. You need a slash in there somewhere, probably like this
$file =~ m#(.*)/([^/]*)$#
It would also make sense to make the directory optional, like so
$file =~ m#(?:(.*)/)?([^/]*)$#
my $directory = $1;
and you would have to write
chdir $directory if $directory

Answer 2

This is what's tripping you up:
$file =~ m#(.*)([^/]*)$#;
Looks like you're trying to split "/tmp/final.pdf" into directory and file. But you don't - that pattern splits you into:
print "F:",$filename,"\n";
print "D:",$directory,"\n";
this output:
F:
D:/tmp/final.pdf
This is why you have the problem - you don't have a filename, so it defaults to using the script name.
I would suggest instead you want:
my ( $directory, $filename ) = ( $file =~ m,(.*/)([\.\w]+)$, );
This gives:
F:final.pdf
D:/tmp/

Answer 3

As has been said, you're suffering from the greedy matching of .* which will eat up the entire string:
$file =~ m{(.*)([^/]*)$};
There are three easy solutions to this
1. Boundary Conditions
As has been stated, you can add a boundary condition that limits how much .* can match:
$file =~ m{(?:(.*)/)?([^/]*)$};
my $dir = $1 // '';
my $filename = $2;
Or this somewhat convoluted lookbehind assertion can also enforce a boundary:
$file =~ m{(.*)(?<![^/])([^/]*)$};
my $dir = $1;
my $filename = $2;
2. Non-greedy matching
However, the simplest regex solution is to use non-greedy matching .*?:
$file =~ m{(.*?)([^/]*)$};
my ($dir, $filename) = ($1, $2);
Basically, anytime you're about to put .* anywhere, check your assumptions. The majority of the time you'll actually want .*? instead.
3. Module for parsing file paths
The bonus option is just to use a module like File::Spec parsing file path information
use File::Spec;
my ($vol, $dirs, $filename) = File::Spec->splitpath( $file );