How do I use a Perl CGI locally without using curl and apache2?

I would like to submit a form to a CGI script localy (w3c-markup-validator), but it is too slow using curl and apache, I want to use this CGI scrip开发者_运维技巧t more than 5,000 times in an another script. and currently it takes more than one hour.

What should I do to give the form directly to the CGI script (I upload a file with curl)?

edit: It seems to be too complicated and time consuming for what I needed, so I waited 1 hour and a half, each time I needed to test my generated xhtml files. In definitive I didn't test any of the answers below, so the question will remain open.

Depending on the details of the script you might be able to create a fake CGI environment using HTTP::Request::AsCGI and then sourcing the CGI script with the "do" operator. But when it comes to speed and maintainability your best bet would be to factor the important part of the script's work into its own module, and rewrite the CGI as a client of that module. That way you don't have to invoke it as a CGI -- the batch job you're talking about now would be just another program using the same module to do the same work, but without CGI or the webserver environment getting in the way.

OK, I looked at the source code for this thing and it is not easy extract the validation stuff from all the rest. So, here is what I would.

First, ditch curl. Starting a new process for each file you want to validate is not a good idea. You are going to need to write a driver script that takes a list of URL's and submits them to your local server running on localhost. In fact, you might later want to parallelize this because there will normally be a bunch of httpd processes alive anyway. Well, I get ahead of myself.

This script can use LWP because all you are doing is submitting some data to the CGI script on localhost and storing/processing results. You do not need full WWW::Mechanize functionality.

As for the validator CGI script, you should configure that as a mod_perl registry script. Make sure you preload all necessary libraries.

This should boost documents processed per second from 1.3 to something more palatable.

CGI is a pretty simple API. All it does is read data either from an environment variable (for GET requests) or from stdin (for POST requests). So all you need is to do is to set up the environment and call the script. See the docs for details.

If the script uses CGI.pm, you can run it from the command line by supplying the '-debug' switch (to CGI.pm, in the use statement.) That will then allow you to send the post variables on stdin. You may have to tweak the script a little to make this work.