CGI::Capture - Meticulously thorough capture and replaying of CGI calls


version 1.15


  # Capture the current CGI to a file, and replay it once created
  use CGI::Capture 'fileupload.dat';
  # Create an object and capture the state
  my $Capture = CGI::Capture->new->capture;
  # Store it in a file and load it back in
  my $second = CGI::Capture->apply('somefile.dat');
  # Apply the CGI call to the current environment


CGI does a terribly bad job of saving CGI calls. CGI::Capture tries to resolve this and save a CGI call in as much painstaking detail as it possibly can.

Because of this, CGI::Capture should work with server logins, cookies, file uploads, strange execution environments, special environment variables, the works.

It does this by capturing a large amount of the perl environment BEFORE itself gets a chance to look at it, and then restores it in the same way.

So in essence, it grabs all of STDIN, %ENV, @INC, and anything else it can think of. The things it can't replicate, it records anyway so that later in the debugger it can ensure that the execution environment is as close as possible to what it captured (and bitch at you about anything you are doing wrong).

This is a huge help when resolving problems such as when a bug won't appear because you aren't debugging the script as the web user and in the same directory.

Using CGI::Capture

The brain-dead way is to use it as a pragma.

Add the following to your web application BEFORE you load in CGI itself.

  use CGI::Capture 'cookiebug.dat';

If the file cookiebug.dat does not exist, CGI::Capture will take a snapshot of all the bits of the environment that matter to a CGI call, and freeze it to the file.

If the file DOES exist however, CGI::Capture will load in the file and replace the current CGI call with the stored one.


The actual captured CGI files are Storable CGI::Capture objects. If you want to use CGI::Capture in an environment where you have CODE references in your @INC path (such as with PAR files), you will need to disable security for Storable by setting $CGI::Capture::DEPARSE to true, which will enable B::Deparse and Eval support for stored objects.

Hand-Crafting CGI Captures

In its default usage, CGI::Capture takes an all or nothing approach, requiring you to capture absolutely every element of a CGI call.

Sometimes you want to be a little more targeted, and for these situations an alternative methodology is provided.

The as_yaml and from_yaml methods allow you to store and retrieve a CGI capture using YAML::Tiny instead of Storable.

Once you have stored the CGI capture as a YAML file, you can hand-edit the capture file, removing any keys you will not want to be restored, keeping only the useful parts.

For example, to create a test file upload or CGI request involving cookies, you could discard everything except for the STDIN section of the capture file, which will then allow you to reuse the capture on other hosts, operating systems, and so on.


In most cases, the above is all you probably need. However, if you want to get more fine-grained control, you can create and manipulate CGI::Capture object directly.


The new only creates a new, empty, capture object.

Because capturing is destructive to some values (STDIN for example) the capture method will capture and then immediately reapply the object, so that the current call can continue.

Returns a CGI::Capture object. Never dies or returns an error, and so can be safely method-chained.

store $filename

This method behaves slightly differently in object and static context.

In object context ( $object->store($filename) ) it stores the captured data to a file via Storable.

In static context ( CGI::Capture->store($filename) ) automatically creates a new capture object, captures the CGI call, and then stores it, all in one hit.

Returns as for Storable::store or dies if there is a problem storing the file. Also dies if it finds a CODE reference in @INC and you have not enabled $CGI::Capture::Deparse.


The retrieve method is used identically to the Storable method of the same name, and wraps it.

Loads in a stored CGI::Capture object from a file.

If the stored object had a CODE ref in it's @INC, you will also need to enable $CGI::Capture::DEPARSE when loading the file.

Returns a new CGI::Capture object, or dies on failure.


To allow for more portable storage and communication of the CGI environment, the as_yaml method can be used to generate a YAML document for the request (generated via YAML::Tiny).

Returns a YAML::Tiny object.


To allow for more portable storage and communication of the CGI environment, the from_yaml method can be used to restore a CGI::Capture object from a YAML::Tiny object.

Returns a new CGI::Capture object, or croaks if passed an invalid parameter.


To allow for more portable storage and communication of the CGI environment, the as_yaml_string method can be used to generate a YAML document for the request (generated via YAML::Tiny).

Returns a YAML document as a string.


To allow for more portable storage and communication of the CGI environment, the from_yaml_string method can be used to restore a CGI::Capture object from a string containing a YAML document.

Returns a new CGI::Capture object, or croaks if the YAML document is invalid.


Again, capture can be used either as an object or static methods

When called as an object method ( $object->capture ) it captures the current CGI call environment into the object, replacing the existing one if needed.

When called as a static method ( CGI::Capture->capture ) it acts as a constructor, creating an object and capturing the CGI call into it before returning it.

In both cases, returns the CGI::Capture object. This method will not die or return an error and can be safely method-chained.

apply [ $filename ]

Again, apply works different when called as an object of static method.

If called as an object method ( $object->apply ) it will take the CGI call the object contains, and apply it to the current environment. Because this works at the environment level, it needs to be done BEFORE attempts to create the CGI object.

The apply method will also check certain values against the current environment. In short, if it can't alter the environment, it won't run unless YOU alter the environment and try again.

These include the real and effective user and group, the OS name, the perl version, and whether Tainting is on or off.

The effect is to really make sure you are replaying the call in your console debugger exactly as it was from the browser, and you aren't accidentally using a different user, a different perl, or are making some other overlooked and hard to debug mistake.

In the future, by request, I may add some options to selectively disable some of the tests. But unless someone asks, I'm leaving all of them on.

In the static context, ( CGI::Capture->apply($file) ) it takes a filename argument, immediately retrieves the CGI call from the object and immediately applies it to the current environment.

In both context, returns true on success or dies on error, or it your testing environment does not match.



Bugs may be submitted through the RT bug tracker (or


Adam Kennedy <>


  • Adam Kennedy <>

  • Karen Etheridge <>


This software is copyright (c) 2004 by Adam Kennedy.

This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.