CGI::Minimal - A lightweight CGI form processing package


 # use CGI::Minimal qw(:no_subprocess_env);
 # use CGI::Minimal qw(:preload);
 use CGI::Minimal;

 my $cgi = CGI::Minimal->new;
 if ($cgi->truncated) {
 my $form_field_value = $cgi->param('some_field_name');


Provides a micro-weight alternative to the module

Rather than attempt to address every possible need of a CGI programmer, it provides the _minimum_ functions needed for CGI such as form decoding (including file upload forms), URL encoding and decoding, HTTP usable date generation (RFC1123 compliant dates) and basic escaping and unescaping of HTMLized text.

The ':preload' use time option is used to force all sub-component modules to load at compile time. It is not required for operation. It is solely a performance optimization for particular configurations. When used, it preloads the 'dehtmlize',' param_mime', 'param_filename', 'date_rfc1123', 'url_decode', 'calling_parms_table' and parameter setting supporting code. Those code sections are normally loaded automatically the first time they are needed.

The ':no_subprocess_env' use time option is used under ModPerl2 to suppress populating the %ENV hash with the usual CGI environmental variables. This is primarily a performance enhancement for some configurations.

The form decoding interface is somewhat compatible with the module. No provision is made for generating HTTP or HTML on your behalf - you are expected to be conversant with how to put together any HTML or HTTP you need.

The parser accepts either '&' or ';' as CGI form field seperators.




both will decode as a=b and c=d

The module supports command line testing of scripts by letting you type a GET style argument followed by pressing the 'enter/return' key when the module is called to decode a form when running a script from the shell.


 bash> ./myscript

Alternatively, you can set the environment variable REQUEST_METHOD to 'GET' and set the environment variable QUERY_STRING to pass your CGI parameters.


bash> REQUEST_METHOD=GET QUERY_STRING='a=b' ./myscript

Performance Hints

If you are using this module as part of a conventional standalone CGI specifically to get a speedup over using, don't 'use warnings', 'use vars' or 'use Carp' in your final production code.

The problem with 'use vars' or 'use warnings' is that (unless you are using Perl 5.8.6 or later) they add on the order of 40 kilobytes of code to your load and you will feel the slowdown. Half of that is from the Carp module (which is used by both the 'vars' and 'warnings' modules.)

So a well written script tuned for performance will start like this:

 #!/usr/bin/perl -Tw

 use strict;
 use CGI::Minimal;

 my $cgi = CGI::Minimal->new;

The ':no_subprocess_env' option is available (use CGI::Minimal qw (:no_subprocess_env);) to suppress population of the %ENV hash with the standard CGI environment values when running under ModPerl2. This improves performance slightly under ModPerl2, but does not affect anything when running as standard CGI or under ModPerl1.

WARNING: Due to the 'singleton' pattern used by CGI::Minimal, the :no_subprocess_env option is persistent between scripts, if you use it in any script, it may affect any ModPerl2 script running on the same Apache instance that uses CGI::Minimal if they run after a script that does use ':no_subprocess_env'.

This would manifest as the %ENV hash as being unexpectedly mostly empty in a script that didn't use ':no_subprocess_env'. The fix for this is to call the CGI::Minimal::subprocess_env class method in any script that requires the %ENV hash to be populated if it is sharing an Apache instance with scripts that use ':no_subprocess_env'.

The ':preload' option is available in CGI::Minimal (use CGI::Minimal qw (:preload);) to compile all sub-components of the module at initial load time. This will slowdown your script unless you are using one or more of the following methods:

 dehtmlize       param_mime    param_filename
 date_rfc1123    url_decode    calling_parms_table

or are using the 'param' method to 'set' values (ie. $cgi->param({'fieldname' => $field_value });).

If you are using one of those methods, then it should provide a slight speedup when being used as a conventional CGI script - but NOT when being used via mod_perl. When invoked via mod_perl, :preload is already implied.

The ultimate performance hint, of course, is use mod_perl (or another persistent execution environment). While you can reach around 66 or so invokations per second on an AMD XP2100+ processor running a simple script using CGI::Minimal, you can exceed 400 per second using mod_perl.

Here are some performance numbers using an extremely simple script using, CGI::Lite, CGI::Deurl,, CGI::Thin, CGI::Simple and CGI::Minimal to read a single passed parameter (a=b) and print it.

Also included are a 'null' Perl script and a 'null' compiled C program which didn't actually read the passed cgi parameter but output the same results as if they had.

Notably, mod_perl outperforms the compiled C program by quite a lot for this simple case. Note also that using the :preload option slowed the CGI::Minimal script down from 66 fetches/second to only 52 fetches/second when used in standard CGI mode.

With 10 parallel fetches repeated as fast as possible for 30 seconds with and without mod_perl (if supported) the tests gave the following results (sorted by speed): (3.05)                 via standard CGI -  16 fetches per second
  CGI::Simple (0.075)           via standard CGI -  20 fetches per second
  CGI::Deurl (1.08)             via standard CGI -  36 fetches per second
  CGI::Thin (0.52)              via standard CGI -  38 fetches per second
  CGI::Lite (2.02)              via standard CGI -  52 fetches per second
  CGI::Minimal (1.16, :preload) via standard CGI -  52 fetches per second
  CGI::Minimal (1.16)           via standard CGI -  66 fetches per second (2.18)             via standard CGI -  71 fetches per second
  null Perl script              via standard CGI - 103 fetches per second
  null C program                via standard CGI - 174 fetches per second
  CGI::Simple (0.075)           via mod_perl     - 381 fetches per second (3.05)                 via mod_perl     - 386 fetches per second
  CGI::Minimal (1.16)           via mod_perl     - 417 fetches per second
  null Perl script              via mod_perl     - 500 fetches per second


 1.29 21 Aug 2007 - Documentation fix to performance hints section. No functional changes.

 1.28 18 Aug 2007 - Improved mod_perl2 handling (patch courtesy of Jeremy Nixon).
                    Added a ':no_subprocess_env' flag to suppress populating
                    the %ENV environment hash. Added a 'subprocess_env'
                    static class method to allow smooth co-existance of
                    ModPerl2 scripts that use ':no_subprocess_env' with ModPerl2
                    scripts that do not on the same server.

 1.27 25 May 2007 - Added example of a command line 'wrapper' script and
                    of using environment variables as an alternate way
                    to test scripts via the command line. Added example
                    for use with FastCGI. Changed behavior for unsupported
                    HTTP methods. The module used to 'croak' for unsupported
                    methods, it now 'carp's instead and treats as a 'GET'
                    (change at suggestion of Roman Mahirov to better support

 1.26 06 Apr 2007 - Added decoding of Javascript/EMCAScript style unicode 
                    escaped (%uxxxx form) parameter data (both to the main
                    'param' method and to the 'url_decode'/'url_encode' methods)
                    at the suggestion of Michael Kröll (the core code for
                    this additional functionality is derived from

                    Added support for ModPerl2.

                    Fixed META.yml problems introduced with 1.25.

                    Changed POD/POD Coverage tests to only execute if specifically requested
                    Added examples directory and scripts

 1.25 20 Apr 2006 - Added 'allow_hybrid_post_get' class method. Tweaked file permissions.
                    Added regression tests for hybrid forms.

 1.24 23 Sep 2005 - Added 'Carp' to install requirements. Extended build tests.
                    Fixed multi-part form decoding bug in handling of degenerate MIME
                    boundaries. Added fatal errors for mis-calling of param_mime
                    and param_filename methods.

 1.23 18 Sep 2005 - Made Test::More optional in build tests. No functional changes.

 1.22 13 Sep 2005 - Changed POD tests to be more friendly to CPANTS.

 1.21 11 Sep 2005 - Fixed pod coverage build test for compatibility
                    with Perl 5.005.

 1.20 11 Sep 2005 - Fixed issue that caused mod_perl to issue
                    'Use of uninitialized value.' warnings.
                    Extended build tests.

 1.19 10 Sep 2005 - Fixed POD Coverage build test error.

 1.18 08 Sep 2005 - Tweaked prerequisite modules lists. Extended regresssion
                    tests to cover more of the code.

 1.17 04 Sep 2005 - More tweaks to regression tests to work around MS-Windows
                    problems with binary file handles under Perl 5.6.1.

                    Added 'Build.PL' support back in. Added POD tests.
                    Minor documentation tweaks. No functional changes.

 1.16 12 Nov 2004 - Added CGI::Simple to the benchmarks table. Tweaked regresssion
                    tests for MS-Windows. Added 'delete' and 'delete_all'
                    methods and regression tests. Added ':preload' flag for
                    preloading all sub-components at module 'use' time.
                    Fixed bug introduced with 1.15 in param value setting.

 1.15 9 Nov 2004  - Added more regression tests. Tweaked url encoder to comply
                    better with RFC2396. Tuned performance some more. Extended
                    benchmarks table to cover more CGI form decoders.

 1.14 16 Oct 2004 - Tuned module load time (about a 40% improvement) and
                    added performance hints to the documentation.

 1.13 28 Sep 2004 - Removed support for Module::Build from module install.

 1.12 25 Sep 2004 - Tweaked the default form parser to accept ';' as a field seperator
                    in addition to '&'. Change suggested by Peter Karman.
                    Eliminated the explicit application/sgml-form-urlencoded support
                    as redundant (it still works, it is just not explicitly different
                    than application/x-www-form-urlencoded support anymore).

                    Adjusted the multipart form parser to be more robust against
                    malformed MIME boundaries and the build tests to work around a
                    bug in Perl 5.8.0 UTF8ness and split.

                    Added documentation of command line script testing behavior.

                    Tightened up the code to reduce size (went from 14.9K to 11K).

                    Removed the 'sgml_safe_mode' redirect code since there
                    was no exposed interface to it anyway.

                    Squashed a bug where the global buffer might fail to initialize
                    correctly for 0 byte POST forms (only impacted  use of the 'raw'
                    method for POST use).

                    Added regression test for form truncation

                    Added LICENSE section to documentation

                    Added Module::Build installation support

 1.11 28 Sep 2003 - Tweaked test script to avoid warnings about
                    opening STDIN filehandle for writing. No functional

 1.10 04 Apr 2003 - Added 'binmode STDIN' on module load to correct for
                    windows defaulting to 7-bit filehandles. Fixed problem where
                    split would fail on unusual MIME boundary strings.
                    Problems noticed by Markus Wichitill.

                    Deferred loading of 'Carp' unless actually needed.

                    Small code cleanups. Removed big disclaimer from
                    .pm and moved in pod to 'DISCLAIMERS' section

                    Added tests

 1.09 19 Mar 2002 - Exposed 'reset_globals' static method for support
                    of non-mod_perl persistent perl execution environments.

 1.08 26 Jul 2001 - Added 'raw' method for obtaining a dump of the raw input buffer data
                    without any parsing. Moved documentation into seperate POD file.

 1.07 01 Dec 2000 - Added capability of taking a GET style parameter string
                         via STDIN if running at the command line.

 1.06 10 Apr 2000 - 'Unfixed' use of 'quotemeta' for multipart
                    boundaries. 'split' is doing the 'wrong thing'
                    according to the specs...

 1.05 03 Apr 2000 - Fixed breakage in 'param;' from 1.04 changes

 1.04 03 Apr 2000 - Added capability to set params via the param() method
                         like '' plus general code cleanup

 1.03 02 Mar 2000 - 'mod_perl' compatibility added

 1.02 09 Jun 1999 - Initial public release.



Creates a new instance of the CGI::Minimal object and decodes any type of form (GET/POST). Only one 'true' object is generated - all subsequent calls return an alias of the one object (a 'Singleton' pattern).


 use CGI::Minimal;

 my $cgi = CGI::Minimal->new;

Called as $cgi->param(); it returns the list of all defined form fields in the same order they appear in the data from the user agent.

Called as $cgi->param($fieldname); it returns the value (or array of values for multiple occurances of the same field name) assigned to that $fieldname. If there is more than one value, the values are returned in the same order they appeared in the data from user agent.


  my (@form_fields) = $cgi->param;

  my (@multi_pick_field) = $cgi->param('pick_field_name');

  my ($form_field_value) = $cgi->param('some_field_name');

You can also use the param method to set param values to new values. These values will be returned by any invokation of a CGI::Minimal object as if they had been found in the original passed data.


    $cgi->param( 'name' => 'Joe Shmoe' );

    $cgi->param({ 'name' => 'Joe Shmoe', 'birth_date' => '06/25/1966' });

    $cgi->param({ 'pick_list' => ['01','05','07'] });

Starting with the 1.12 version, CGI::Minimal accepts the ';' character as a form field seperator in addition to the '&'.

IOW: a=b;c=d will decode as well as a=b&c=d in form submissions. This is to provide SGML/XML/XHTML compatibility for those needing it. Since ';' has always been escaped to %3e by the url_encode method, this should only affect people who are already doing the wrong thing with their forms by not escaping their data correctly. This is also compatible with what the module does.


This is functionally equivalent to the 'delete_all' method. It clears all currently set param values. It is not the same as the static method reset_globals because it does not reset the raw, max_read_size, truncated states or reintialize the form decoding configuration for reuse.

It just deletes all the parameters.



This allows you to delete specified parameters from the CGI::Minimal object.

Ex. $cgi->delete('fieldname', 'another');


Called as $cgi->param_mime(); it returns the list of all defined form fields in the same order they appear in the data from the user agent.

Called as $cgi->param_mime($fieldname); it returns the MIME type (or array of MIME types for multiple occurances of the same field name) assigned to that $fieldname. If there is more than one value, the values are returned in the same order they appeared in the data from user agent.

This is only meaningful when doing Form Based File Uploads and should probably not be trusted even then since it depends on the _browser_ correctly identifying what it is sending.


Called as $cgi->param_filename(); it returns the list of all defined form fields in the same order they appear in the data from the user agent.

Called as $cgi->param_filename($fieldname); it returns the file name (or array of file names for multiple occurances of the same field name) assigned to that $fieldname. If there is more than one value, the values are returned in the same order they appeared in the data from user agent.

This is only meaningful when doing Form Based File Uploads.


Returns a copy of the raw form data. Returns 'undef' if the form has not been parsed yet.


Takes a unix time tick value and returns a RFC1123 compliant date as a formatted text string suitable for direct use in Expires, Last-Modified or other HTTP headers (with the exception of 'Set-Cookie', which requires a different format not generated here. See 'CGI::Cookie' for cookie generation).


 print "Expires: ",$cgi->date_rfc1123(time + 3600),"\015\012";

Returns a formatted HTML table containing all the form and environment variables for debugging purposes


  print $cgi->calling_parms_table;

Returns URL encoding of input string (URL unsafe codes are escaped to %xx form)


 my $url_encoded_string = $cgi->url_encode($string);

Returns URL *decoding* of input string (%xx and %uxxxx substitutions are decoded to their actual values).


 my $url_decoded_string = $cgi->url_decode($string);

Returns HTML 'safe' encoding of input string. Replaces &,>,< and " with their named entity codes (&amp, &gt; &lt; and &quot;)


 my $html_escaped_string = $cgi->htmlize($string);

Undoes basic HTML encoding of input string. Replaces &amp;, &gt;, &lt; and &quot; named entity codes with their actual values. NOT a general purpose entity decoder.


Returns '1' if the read form was shorter than the Content-Length that was specified by the submitting user agent (ie the data from a form uploaded by a web browser was cut off before all the data was received).

Returns '0' if the form was NOT truncated.


  use CGI::Minimal;

  my $cgi = CGI::Minimal->new;
  if ($cgi->truncated) {
  } else {

'truncated' will also return '1' if the form length received would have exceeded the set 'max_read_length'.



Sets the maximum number of bytes/octets that the CGI decoder will accept. By default, 1000000 bytes/octets. This must be called *BEFORE* calling 'new' for the first time for it to actually affect form decoding.


  use CGI::Minimal;

  my $cgi = CGI::Minimal->new;

The HTTP standard specifies that POST parameters are passed in the body of the HTTP request while GET parameters are passed as part of the URL. The two are not supposed to be mixed, and the semantics of doing so are undefined, but many people have done so anyway.

By default, CGI::Minimal does not mix the two.

If you wish to do so (with a resulting speed penalty for processing 'POST' forms since it has to also process the GET query string) you can do so by setting allow hybrid post/get via the allow_hybrid_post_get static method before calling 'new' for the first time:


  use CGI::Minimal;

  my $cgi = CGI::Minimal->new;

CGI::Minimal decodes the POST parameters first, and appends the GET parameters after them.

Note: Setting allow hybrid post/get on a POST does not append the GET query string (if any) to the POST data returned by the 'raw' method: It continues to return only the raw POST message body for a POST request and the raw GET query string for a GET (or HEAD) request.


Sets the CGI::Minimal object to it's initial state (the state it is in before calling 'new' for the first time in a CGI interaction).

This static method not needed if you are running CGI::Minimal either as a traditional CGI script or under mod_perl. In other persistent execution environments you need to call this static method _before_ calling ->new or the module will not 'forget' the data parameters passed from the last time it was executed.

FCGI Example:

 #!/usr/bin/perl -Tw

 use strict;
 use FCGI;
 use CGI::Minimal;

 my $request = FCGI::Request();
 while($request->Accept() >= 0) {
    my $value = CGI::Minimal->new->param('a');
    print "Status: 200 OK\015\012Content-Type: text/plain\015\012\015\012a = $value\n";

Note: This method resets the the 'max_read_size' back to its default of 1_000_000 octets ('bytes'); so if you want to use a different size, you need to restore it after you do the 'reset_globals' call.


 use CGI::Minimal;

 my $cgi = CGI::Minimal->new;

Under ModPerl2, this imports the CGI environment variables into the %ENV hash. Under either a standard CGI invokation or ModPerl1, it does nothing. It is intended for use in scripts sharing a ModPerl2 Apache instance with scripts that use the ':no_subprocess_env' flag.

Because of the 'singleton' pattern used by CGI::Minimal, if one ModPerl2 script sets the ':no_subprocess_env' flag, other scripts may be affected by it (which would manifest as mysteriously missing %ENV hash values).

To prevent that from happening, add


to that scripts that need %ENV to be populated if they share an Apache instance with other scripts that DO use the :no_subprocess_env flag.


 use CGI::Minimal

 my $cgi = CGI::Minimal->new;

The method has no effect on regular CGI or ModPerl1 scripts and can be safely invoked in them without harm.


None known.


To be determined. ^_^


Benjamin Franz <>


Version 1.29 21 Aug 2007


Copyright (c) Benjamin Franz. All rights reserved.


This program is free software; you can redistribute it and/or modify it under the same terms and conditions as Perl itself.

This means that you can, at your option, redistribute it and/or modify it under either the terms the GNU Public License (GPL) version 1 or later, or under the Perl Artistic License.




Use of this software in any way or in any form, source or binary, is not allowed in any country which prohibits disclaimers of any implied warranties of merchantability or fitness for a particular purpose or any disclaimers of a similar nature.




1 POD Error

The following errors were encountered while parsing the POD:

Around line 184:

Non-ASCII character seen before =encoding in 'Kröll'. Assuming UTF-8