NAME

CGI::Compile - Compile .cgi scripts to a code reference like ModPerl::Registry

SYNOPSIS

use CGI::Compile;
my $sub = CGI::Compile->compile("/path/to/script.cgi");

DESCRIPTION

CGI::Compile is an utility to compile CGI scripts into a code reference that can run many times on its own namespace, as long as the script is ready to run on a persistent environment.

NOTE: for best results, load CGI::Compile before any modules used by your CGIs.

RUN ON PSGI

Combined with CGI::Emulate::PSGI, your CGI script can be turned into a persistent PSGI application like:

use CGI::Emulate::PSGI;
use CGI::Compile;

my $cgi_script = "/path/to/foo.cgi";
my $sub = CGI::Compile->compile($cgi_script);
my $app = CGI::Emulate::PSGI->handler($sub);

# $app is a PSGI application

CAVEATS

If your CGI script has a subroutine that references the lexical scope variable outside the subroutine, you'll see warnings such as:

Variable "$q" is not available at ...
Variable "$counter" will not stay shared at ...

This is due to the way this module compiles the whole script into a big sub. To solve this, you have to update your code to pass around the lexical variables, or replace my with our. See also http://perl.apache.org/docs/1.0/guide/porting.html#The_First_Mystery for more details.

METHODS

new

Does not need to be called, you only need to call it if you want to set your own namespace_root for the generated packages into which the CGIs are compiled into.

Otherwise you can just call "compile" as a class method and the object will be instantiated with a namespace_root of CGI::Compile::ROOT.

You can also set return_exit_val, see "RETURN CODE" for details.

Example:

my $compiler = CGI::Compile->new(namespace_root => 'My::CGIs');
my $cgi      = $compiler->compile('/var/www/cgi-bin/my.cgi');

compile

Takes either a path to a perl CGI script or a source code and some other optional parameters and wraps it into a coderef for execution.

Can be called as either a class or instance method, see "new" above.

Parameters:

  • $cgi_script

    Path to perl CGI script file or a scalar reference that contains the source code of CGI script, required.

  • $package

    Optional, package to install the script into, defaults to the path parts of the script joined with _, and all special characters converted to _%2x, prepended with CGI::Compile::ROOT::.

    E.g.:

    /var/www/cgi-bin/foo.cgi

    becomes:

    CGI::Compile::ROOT::var_www_cgi_2dbin_foo_2ecgi

Returns:

  • $coderef

    $cgi_script or $$code compiled to coderef.

SCRIPT ENVIRONMENT

ARGUMENTS

Things like the query string and form data should generally be in the appropriate environment variables that things like CGI expect.

You can also pass arguments to the generated coderef, they will be locally aliased to @_ and @ARGV.

BEGIN and END blocks

BEGIN blocks are called once when the script is compiled. END blocks are called when the Perl interpreter is unloaded.

This may cause surprising effects. Suppose, for instance, a script that runs in a forking web server and is loaded in the parent process. END blocks will be called once for each worker process and another time for the parent process while BEGIN blocks are called only by the parent process.

%SIG

The %SIG hash is preserved meaning the script can change signal handlers at will. The next invocation gets a pristine %SIG again.

exit and exceptions

Calls to exit are intercepted and converted into exceptions. When the script calls exit 19 and exception is thrown and $@ contains a reference pointing to the array

["EXIT\n", 19]

Naturally, "$^S" in perlvar (exceptions being caught) is always true during script runtime.

If you really want to exit the process call CORE::exit or set $CGI::Compile::USE_REAL_EXIT to true before calling exit:

$CGI::Compile::USE_REAL_EXIT = 1;
exit 19;

Other exceptions are propagated out of the generated coderef. The coderef's caller is responsible to catch them or the process will exit.

Return Code

The generated coderef's exit value is either the parameter that was passed to exit or the value of the last statement of the script. The return code is converted into an integer.

On a 0 exit, the coderef will return 0.

On an explicit non-zero exit, by default an exception will be thrown of the form:

exited nonzero: <n>

where n is the exit value.

This only happens for an actual call to "exit" in perfunc, not if the last statement value is non-zero, which will just be returned from the coderef.

If you would prefer that explicit non-zero exit values are returned, rather than thrown, pass:

return_exit_val => 1

in your call to "new".

Alternately, you can change this behavior globally by setting:

$CGI::Compile::RETURN_EXIT_VAL = 1;

Current Working Directory

If CGI::Compile->compile was passed a script file, the script's directory becomes the current working directory during the runtime of the script.

NOTE: to be able to switch back to the original directory, the compiled coderef must establish the current working directory. This operation may cause an additional flush operation on file handles.

STDIN and STDOUT

These file handles are not touched by CGI::Compile.

The DATA file handle

If the script reads from the DATA file handle, it reads the __DATA__ section provided by the script just as a normal script would do. Note, however, that the file handle is a memory handle. So, fileno DATA will return -1.

CGI.pm integration

If the subroutine CGI::initialize_globals is defined at script runtime, it is called first thing by the compiled coderef.

AUTHOR

Tatsuhiko Miyagawa <miyagawa@bulknews.net>

CONTRIBUTORS

Rafael Kitover <rkitover@cpan.org>

Hans Dieter Pearcey <hdp@cpan.org>

kocoureasy <igor.bujna@post.cz>

Torsten Förtsch <torsten.foertsch@gmx.net>

COPYRIGHT & LICENSE

Copyright (c) 2009 Tatsuhiko Miyagawa <miyagawa@bulknews.net>

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

SEE ALSO

ModPerl::RegistryCooker CGI::Emulate::PSGI