In the docs it says filenames only

If $input is an array reference, each element in the array must be a filename. The input data will be read from each file in turn. The complete array will be walked to ensure that it only contains valid filenames before any data is compressed.

IO::Compress::Zip cannot write directories (without jumping through a lot of hoops) by design.

I do have another module, called Archive::Zip::SimpleZip (it actually uses IO::Compress::Zip under the hood), that can create directories.

Here is a example of how to use it

use strict ;
use warnings;
use Archive::Zip::SimpleZip;

my @dirs = ("WEB-INF");

my $zip = Archive::Zip::SimpleZip->new("test.zip");
$zip->add($_) for @dirs;
$zip->close();
[download]

and here is what the commandline unzip this about the zip file

$ unzip -l test.zip 
Archive:  test.zip
  Length      Date    Time    Name
---------  ---------- -----   ----
        0  02-03-2016 12:58   WEB-INF/
---------                     -------
        0                     1 file
[download]

But I guess I wasn't clear enough.

Actually, I think that I understood you correctly, but I probably failed to properly describe what I tested.

Here's the file and directory layout of what I tested (and the contents of test.pl is the code from my previous post):

C:\test\test.pl
C:\test\test_dir\file1.txt
C:\test\test_dir\file2.txt
[download]

After running perl test.pl, I got a new file: C:\test\test.zip. Here's the contents of test.zip:

test_dir\file1.txt
test_dir\file2.txt
[download]

As you can see, the file and directory structure was fully and correctly added into the new zip file. And a quick check shows that the two text files' modification date/time stamp is preserved in the new zip file too. I believe that is the basic idea of what you are trying to do. However, I suspect that you'll need to make modifications to the code that I posted in order to fully do what you're trying to do.