CGI.pm woes with multipart attachments

zentara has asked for the wisdom of the Perl Monks concerning the following question:

Hi, recently I posted a snippet to allow multipart attachment
downloads. I generated my own headers and it works fine with
linux apache and Mozilla. Merlyn reponded, that I could also
do it with CGI.pm. Well at the time, I didn't have the latest version
of CGI.pm, which supported the sub multipart_final, so I just
shoved the script to my code collection. I recently upgraded to
the latest CGI.pm and tried Merlyn's script again, and it just
refuses to run for me. I am asking for "higher guidance" now.
My working code follows

#!/usr/bin/perl
use warnings;
use strict;

my $boundary_string = "\n" . "--End" . "\n";
my $end_of_data = "\n" . "--End--" . "\n";

my @file_list = ("test.tgz","test1.tgz","test2.tgz");

print<<EOH;
Content-type: multipart/x-mixed-replace\;boundary=End

EOH

foreach my $file (@file_list){
     &amp;amp;send_file ($file);
     print $boundary_string;
}
print $end_of_data;
exit(0);

sub send_file {
my $file = $_[0];
if (open (FILE, "< $file")) {
print<<EOF; 
Content-type: application/octet-stream 
Content-Disposition: attachment\; filename=$file 
 
EOF
binmode FILE;
print <FILE>;
close (FILE);
}else{print "Cannot open file $file!"}
}
[download]

############################################
merlyn's suggestion follows below:

#!/usr/bin/perl
use CGI qw(:push);
@ARGV = ("test.tgz","test1.tgz","test2.tgz");
$/ = undef;

print multipart_init();
while (<>) {
print multipart_start("application/octet-stream");
print $_;
print multipart_end();
}
print multipart_final();
[download]

########################################
When I run Merlyn's code, the filename is always
nph-download-multi, and I'm prompted to save that
filename 6 times instead of 3, and the name is always
nph-download-multi, instead of test.tgz, test1.tgz, etc.
Now I did try to add a "-filename=>$_" to the code to
get the filename to come thru, but to no avail. Also the resultant
download is 0 bytes. I've tried every combination of header
that I could think of, but no luck. So would anyone be kind
enough to post a multipart-download script , based on CGI.pm
which actually works, so I can see what I'm doing wrong?
Thanks

Comment on CGI.pm woes with multipart attachments Select or Download Code

Replies are listed 'Best First'.

Re: CGI.pm woes with multipart attachments
by sedhed (Scribe) on Jul 25, 2002 at 21:53 UTC

Perhaps a piece of the puzzle: Your magic line Content-Disposition: attachment\; filename=$file is what suggests the filename for the browser. CGI's multipart_start() function does not print this. However, you should be able to pass it yourself:

print multipart_start(
    -TYPE=>'application/octet-stream',
    '-Content-Disposition'=>"attachment\; filename=test.tgz", # Have t
+o quote the directive here
);
[download]

An aside: I don't know how, offhand, to determine the current filename using merlyn's code, other than a kludge like $filename = $ARGV[$count++]; since $_ is the contents of each file, not the file's name.

Update: Sorry if this is obvious: Don't forget to try running it from a command-line to debug. That will let you see the exact headers and file contents the script is printing.

Cheers!

[reply]
[d/l]
[select]


Think about Loose Coupling
	PerlMonks