Compile your cgi into a subroutine with CGI::Compile, wrap it with CGI::Emulate::PSGI, then install LWP::Protocol::PSGI, so you can talk to your cgi with WWW::Mechanize

Or maybe just use Test::WWW::Mechanize::CGI, it uses HTTP::Request::AsCGI

Thanks for the suggestion, however, I think I am not going to trigger any http mechanism at all (for the sake of a slow notebook with 10K+ batch requests). I just wanted to write into STDIN (if any chance or alternative approach). Just want my main script get the form like it's from server self.

Yes, I think both parts of my answer cover both situations :) 10K+ really ain't that many :)

#!/usr/bin/perl -wT
use strict;

my $buffer;
read (STDIN, $buffer, $ENV{'CONTENT_LENGTH'});
print "Content-type: text/plain\n\n";
print $buffer;
[download]

#!/usr/bin/perl

# Description:  Takes all CGI parameters (both GET and POST data) 
# and builds a query string to send as a GET request.

use CGI::Simple; # or use CGI;
use URI::Escape;

my $CGI = CGI::Simple->new; # or CGI->new;

my $query = join(
  ';', map {
     uri_escape($_) . '=' . uri_escape($CGI->param($_))
  } $CGI->param
);

# now we can send a GET rquest somewhere
$CGI->redirect('http://my.server.com/script.pl?' . $query);
[download]

Thanks zentara, that's actually what I am up to so far, I alter a POST request to a GET request, however, like the code I stated at the last paragraph, this won't work when there's a single script that handle GET and POST in a total different manner, provided I want to leave the old codes as is. So I just hope to find a way to write into the STDIN, so the read() later can just implement like the data really comes from the web server self. Is it possible to write into STDIN anyway ?

'bar's STDIN is whatever 'foo' spits to STDOUT. So if you can control how your code is executed from the shell, you can control its STDIN. For example ...

for i in `seq 1 10`; do echo $i|./myscript.pl; done
[download]

Hey, this one is interesting and seem very likely what I am looking for... but please have a look for what I am planning to do, and could you advise can I do this inside the code rather than pipe it from the shell?

$ENV{REQUEST_METHOD} = "POST";

open F, "testplan.txt" or die $!;
while ( chomp ( my $data = <F> ) ) {

  ### DO ANYTHING TO MAKE $data BECOME STDIN ###  

  # the main() script that reads and handle the form
  # print final result
  # next test plan data
}
close F;
[download]

You could close STDIN and re-open it as a normal filehandle:

close(STDIN);
open(STDIN, 'foo');
print $_ foreach(<STDIN>);
[download]

The documentation for 'open' shows how to save and restore filehandles, and some other dirty tricks you can play.

Generically speaking, you can feed the STDIN of a child process like so:

open my $fh, '|command' or die "Can't run command: $!";
print $fh "input for command\n";
print $fh "more input for command\n";
close $fh;
[download]

More specific to your situation, years ago, I wrote a simple script to copy STDIN to a file, then set up Lighttp to run that script as a CGI. Then I opened a HTML file, with the form I needed to fill, in a web browser, filled in the form and submitted. I then edited (a copy of) the resulting file to create a template. with thta, I could use a script to test my target CGI by inserting various values into the template and feeding the result to the CGI script similarly to my code snippet above.