NAME
CGI::Portable - Framework for server-generic web apps
DEPENDENCIES
Perl Version
5.004Standard Modules
I<none>Nonstandard Modules
CGI::Portable::Files 0.461
CGI::Portable::Request 0.471
CGI::Portable::Response 0.461SYNOPSIS
Content of thin shell "startup_cgi.pl" for CGI or Apache::Registry env:
#!/usr/bin/perl
use strict;
use warnings;
require CGI::Portable;
my $globals = CGI::Portable->new();
use Cwd;
$globals->file_path_root( cwd() ); # let us default to current working directory
$globals->file_path_delimiter( $^O=~/Mac/i ? ":" : $^O=~/Win/i ? "\\" : "/" );
$globals->set_prefs( 'config.pl' );
$globals->current_user_path_level( 1 );
require CGI::Portable::AdapterCGI;
my $io = CGI::Portable::AdapterCGI->new();
$io->fetch_user_input( $globals );
$globals->call_component( 'Aardvark' );
$io->send_user_output( $globals );
1;Content of thin shell "startup_socket.pl" for IO::Socket::INET:
#!/usr/bin/perl
use strict;
use warnings;
print "[Server $0 starting up]\n";
require CGI::Portable;
my $globals = CGI::Portable->new();
use Cwd;
$globals->file_path_root( cwd() ); # let us default to current working directory
$globals->file_path_delimiter( $^O=~/Mac/i ? ":" : $^O=~/Win/i ? "\\" : "/" );
$globals->set_prefs( 'config.pl' );
$globals->current_user_path_level( 1 );
require CGI::Portable::AdapterSocket;
my $io = CGI::Portable::AdapterSocket->new();
use IO::Socket;
my $server = IO::Socket::INET->new(
Listen => SOMAXCONN,
LocalAddr => '127.0.0.1',
LocalPort => 1984,
Proto => 'tcp'
);
die "[Error: can't setup server $0]" unless $server;
print "[Server $0 accepting clients]\n";
while( my $client = $server->accept() ) {
printf "%s: [Connect from %s]\n", scalar localtime, $client->peerhost;
my $content = $globals->make_new_context();
$io->fetch_user_input( $content, $client );
$content->call_component( 'Aardvark' );
$io->send_user_output( $content, $client );
close $client;
printf "%s http://%s:%s%s %s\n", $content->request_method,
$content->server_domain, $content->server_port,
$content->user_path_string, $content->http_status_code;
}
1;Content of settings file "config.pl"
my $rh_prefs = {
title => 'Welcome to Aardvark',
credits => '<p>This program copyright 2001 Darren Duncan.</p>',
screens => {
one => {
'link' => 'Fill Out A Form',
mod_name => 'Tiger',
mod_prefs => {
field_defs => [
{
visible_title => "What's your name?",
type => 'textfield',
name => 'name',
}, {
visible_title => "What's the combination?",
type => 'checkbox_group',
name => 'words',
'values' => ['eenie', 'meenie', 'minie', 'moe'],
default => ['eenie', 'minie'],
rows => 2,
}, {
visible_title => "What's your favorite colour?",
type => 'popup_menu',
name => 'color',
'values' => ['red', 'green', 'blue', 'chartreuse'],
}, {
type => 'submit',
},
],
},
},
two => {
'link' => 'Fly Away',
mod_name => 'Owl',
mod_prefs => {
fly_to => 'http://www.perl.com',
},
},
three => {
'link' => 'Don\'t Go Here',
mod_name => 'Camel',
mod_subdir => 'files',
mod_prefs => {
priv => 'private.txt',
prot => 'protected.txt',
publ => 'public.txt',
},
},
four => {
'link' => 'Look At Some Files',
mod_name => 'Panda',
mod_prefs => {
food => 'plants',
color => 'black and white',
size => 'medium',
files => [qw( priv prot publ )],
file_reader => '/three',
},
},
},
};Content of fat main program component "Aardvark.pm"
This module acts sort of like CGI::Portable::AppMultiScreen.
package Aardvark;
use strict;
use warnings;
use CGI::Portable;
sub main {
my ($class, $globals) = @_;
my $users_choice = $globals->current_user_path_element();
my $rh_screens = $globals->pref( 'screens' );
if( my $rh_screen = $rh_screens->{$users_choice} ) {
my $inner = $globals->make_new_context();
$inner->inc_user_path_level();
$inner->navigate_url_path( $users_choice );
$inner->navigate_file_path( $rh_screen->{mod_subdir} );
$inner->set_prefs( $rh_screen->{mod_prefs} );
$inner->call_component( $rh_screen->{mod_name} );
$globals->take_context_output( $inner );
} else {
$globals->set_page_body( "<p>Please choose a screen to view.</p>" );
foreach my $key (keys %{$rh_screens}) {
my $label = $rh_screens->{$key}->{link};
my $url = $globals->url_as_string( $key );
$globals->append_page_body( "<br /><a href=\"$url\">$label</a>" );
}
}
$globals->page_title( $globals->pref( 'title' ) );
$globals->prepend_page_body( "<h1>".$globals->page_title()."</h1>\n" );
$globals->append_page_body( $globals->pref( 'credits' ) );
}
1;Content of component module "Tiger.pm"
This module acts sort of like DemoMailForm without the emailing.
package Tiger;
use strict;
use warnings;
use CGI::Portable;
use HTML::FormTemplate;
sub main {
my ($class, $globals) = @_;
my $ra_field_defs = $globals->resolve_prefs_node_to_array(
$globals->pref( 'field_defs' ) );
if( $globals->get_error() ) {
$globals->set_page_body(
"Sorry I can not do that form thing now because we are missing ",
"critical settings that say what the questions are.",
"Reason: ", $globals->get_error(),
);
$globals->add_no_error();
return( 0 );
}
my $form = HTML::FormTemplate->new();
$form->form_submit_url( $globals->recall_url() );
$form->field_definitions( $ra_field_defs );
$form->user_input( $globals->user_post() );
$globals->set_page_body(
'<h1>Here Are Some Questions</h1>',
$form->make_html_input_form( 1 ),
'<hr />',
'<h1>Answers From Last Time If Any</h1>',
$form->new_form() ? '' : $form->make_html_input_echo( 1 ),
);
}
1;Content of component module "Owl.pm"
This module acts sort of like DemoRedirect.
package Owl;
use strict;
use warnings;
use CGI::Portable;
sub main {
my ($class, $globals) = @_;
my $url = $globals->pref( 'fly_to' );
$globals->http_status_code( '301 Moved' );
$globals->http_redirect_url( $url );
}
1;Content of component module "Camel.pm"
This module acts sort of like DemoStatic.
package Camel;
use strict;
use warnings;
use CGI::Portable;
sub main {
my ($class, $globals) = @_;
my $users_choice = $globals->current_user_path_element();
my $filename = $globals->pref( $users_choice );
my $filepath = $globals->physical_filename( $filename );
SWITCH: {
$globals->add_no_error();
open( FH, $filepath ) or do {
$globals->add_virtual_filename_error( 'open', $filename );
last SWITCH;
};
local $/ = undef;
defined( my $file_content = <FH> ) or do {
$globals->add_virtual_filename_error( "read from", $filename );
last SWITCH;
};
close( FH ) or do {
$globals->add_virtual_filename_error( "close", $filename );
last SWITCH;
};
$globals->set_page_body( $file_content );
}
if( $globals->get_error() ) {
$globals->append_page_body(
"Can't show requested screen: ".$globals->get_error() );
$globals->add_no_error();
}
}
1;Content of component module "Panda.pm"
This module acts sort of like nothing I've ever seen.
package Panda;
use strict;
use warnings;
use CGI::Portable;
sub main {
my ($class, $globals) = @_;
$globals->set_page_body( <<__endquote );
<p>Food: @{[$globals->pref( 'food' )]}
<br />Color: @{[$globals->pref( 'color' )]}
<br />Size: @{[$globals->pref( 'size' )]}</p>
<p>Now let's look at some files; take your pick:
__endquote
$globals->navigate_url_path( $globals->pref( 'file_reader' ) );
foreach my $frag (@{$globals->pref( 'files' )}) {
my $url = $globals->url_as_string( $frag );
$globals->append_page_body( "<br /><a href=\"$url\">$frag</a>" );
}
$globals->append_page_body( "</p>" );
}
1;DESCRIPTION
The CGI::Portable class is a framework intended to support complex web applications that are easily portable across servers because common environment-specific details are abstracted away, including the file system type, the web server type, and your project's location in the file system or uri hierarchy.
Also abstracted away are details related to how users of your applications arrange instance config/preferences data across single or multiple files, so they get more flexability in how to use your application without you writing the code to support it. So your apps are easier to make data-controlled.
Application cores would use CGI::Portable as an interface to the server they are running under, where they receive user input through it and they return a response (HTML page or other data type) to the user through it. Since CGI::Portable should be able to express all of their user input or output needs, your application cores should run well under CGI or mod_perl or IIS or a Perl-based server or a command line without having code that supports each type's individual needs.
That said, CGI::Portable doesn't contain any user input/output code of its own, but allows you to use whatever platform-specific code or modules you wish between it and the actual server. By using my module as an abstraction layer, your own program core doesn't need to know which platform-specific code it is talking to.
As a logical extension to the interfacing functionality, CGI::Portable makes it easier for you to divide your application into autonomous components, each of which acts like it is its own application core with user input and instance config data provided to it and a recepticle for its user output provided. This module would be an interface between the components.
This class inherits most of its functionality from four other modules that were created with that intent, although each can be used independantly as well:
- CGI::Portable::Errors
- CGI::Portable::Files
- CGI::Portable::Request
- CGI::Portable::ResponseEach module has complete POD for functionality it implements, with conceptual overviews and method descriptions. Please read them in order to have the best understanding of what CGI::Portable can do.
SIMILAR MODULES
Based on the above, you could conceivably say CGI::Portable has similarities to these modules: CGI::Screen, CGI::MxScreen, CGI::Application, CGI::BuildPage, CGI::Response, HTML::Mason, CGI, and others.
To start with, all of the above modules do one or more of: storing and providing access to user input, helping to organize access to multiple user screens or application modes, collecting and storing output for the user, and so on.
Some ways that the modules are different from mine are: level of complexity, because my module is simpler than HTML::Mason and CGI::MxScreen and CGI, but it is more complex and/or comprehensive than the others; functionality, because it takes portability between servers to a new level by being agnostic on both ends, where the other solutions are all/mostly tied to specific server types since they do the I/O by themselves; my module also does filesystem translation and some settings management, and I don't think any of the others do; I have built-in functionality for organizing user screens hierarchically, called user_path/url_path (in/out equivalents); I keep query params and post params separate whereas most of the others use CGI.pm which combines them together; more differences.
YES, THIS MODULE DOES IMAGES
Just in case you were thinking that this module does plain html only and is no good for image-making applications, let me remind you that, yes, CGI::Portable can map urls to, store, and output any type of file, including pictures and other binary types.
To illustrate this, I have provided the "image" demo consisting of an html page containing a PNG graphic, both of which are generated by the same script. (You will need to have GD installed to see the picture, though.)
Besides that, this module has explicit support for the likes of cascading style sheets (css) and complete multi-frame documents in one script as well, which are normally just used in graphical environments.
So while a few critics have pointed out the fact that my own websites, which use this module, don't have graphics, then that is purely my own preference as a way to make them load faster and use less bandwidth, not due to any lack of the ability to use pictures.
A DIFFERENT OVERVIEW
This class is designed primarily as a data structure that intermediates between your large central program logic and the small shell part of your code that knows anything specific about your environment. The way that this works is that the shell code instantiates an CGI::Portable object and stores any valid user input in it, gathered from the appropriate places in the current environment. Then the central program is started and given the CGI::Portable object, from which it takes stored user input and performs whatever tasks it needs to. The central program stores its user output in the same CGI::Portable object and then quits. Finally, the shell code takes the stored user output from the CGI::Portable object and does whatever is necessary to send it to the user. Similarly, your thin shell code knows where to get the instance-specific file system and stored program settings data, which it gives to the CGI::Portable object along with the user input.
Here is a diagram:
YOUR THIN CGI::Portable YOUR FAT "CORE"
USER <----> "MAIN" CONFIG, <----> INTERFACE LAYER <----> PROGRAM LOGIC
I/O SHELL FRAMEWORK FUNCTIONALITY
(may be portable) (portable) (portable)This class does not gather any user input or send any user input by itself, but expects your thin program instance shell to do that. The rationale is both for keeping this class simpler and for keeping it compatible with all types of web servers instead of just the ones it knows about. So it works equally well with CGI under any server or mod_perl or when your Perl is its own web server or when you are debugging on the command line.
Because your program core uses this class to communicate with its "superior", it can be written the same way regardless of what platform it is running on. The interface that it needs to written to is consistent across platforms. An analogy to this is that the core always plays in the same sandbox and that environment is all it knows; you can move the sandbox anywhere you want and its occupant doesn't have to be any the wiser to how the outside world had changed.
From there, it is a small step to breaking your program core into reusable components and using CGI::Portable as an interface between them. Each component exists in its own sandbox and acts like it is its own core program, with its own task to produce an html page or other http response, and with its own set of user input and program settings to tell it how to do its job. Depending on your needs, each "component" instance could very well be its own complete application, or it would in fact be a subcontractee of another one. In the latter case, the "subcontractor" component may have other components do a part of its own task, and then assemble a derivative work as its own output.
When one component wants another to do work for it, the first one instantiates a new CGI::Portable object which it can pass on any user input or settings data that it wishes, and then provides this to the second component; the second one never has to know where its CGI::Portable object it has came from, but that everything it needs to know for its work is right there. This class provides convenience methods like make_new_context() to simplify this task by making a partial clone that replicates input but not output data.
Due to the way CGI::Portable stores program settings and other input/output data, it lends itself well to supporting data-driven applications. That is, your application components can be greatly customizable as to their function by simply providing instances of them with different setup data. If any component is so designed, its own config instructions can detail which other components it subcontracts, as well as what operating contexts it sets up for them. This results in a large variety of functionality from just a small set of components.
Another function that CGI::Portable provides for component management is that there is limited protection for components that are not properly designed to be kept from harming other ones. You see, any components designed a certain way can be invoked by CGI::Portable itself at the request of another component. This internal call is wrapped in an eval block such that if a component fails to compile or has a run-time exception, this class will log an error to the effect and the component that called it continues to run. Also, called components get a different CGI::Portable object than the parent, so that if they mess around with the stored input/output then the parent component's own data isn't lost. It is the parent's own choice as to which output of its child that it decides to copy back into its own output, with or without further processing.
Note that the term "components" above suggests that each one is structured as a Perl 5 module and is called like one; the module should have a method called main() that takes an CGI::Portable object as its argument and has the dispatch code for that component. Of course, it is up to you.
SYNTAX
This class does not export any functions or methods, so you need to call them using object notation. This means using Class->function() for functions and $object->method() for methods. If you are inheriting this class for your own modules, then that often means something like $self->method().
CONSTRUCTOR FUNCTIONS AND METHODS
These functions and methods are involved in making new CGI::Portable objects.
new([ FILE_ROOT[, FILE_DELIM[, PREFS]] ])
This function creates a new CGI::Portable (or subclass) object and returns it. All of the method arguments are passed to initialize() as is; please see the POD for that method for an explanation of them.
initialize([ FILE_ROOT[, FILE_DELIM[, PREFS]] ])
This method is used by new() to set the initial properties of objects that it creates. The optional 3 arguments are used in turn to set the properties accessed by these methods: file_path_root(), file_path_delimiter(), set_prefs().
clone([ CLONE ])
This method initializes a new object to have all of the same properties of the current object and returns it. This new object can be provided in the optional argument CLONE (if CLONE is an object of the same class as the current object); otherwise, a brand new object of the current class is used. Only object properties recognized by CGI::Portable are set in the clone; other properties are not changed.
METHODS FOR CONTEXT SWITCHING
These methods are designed to facilitate easy modularity of your application into multiple components by providing context switching functions for the parent component in a relationship. While you could still use this class effectively without using them, they are available for your convenience.
make_new_context([ CONTEXT ])
This method initializes a new object of the current class and returns it. This new object has some of the current object's properties, namely the "input" properties, but lacks others, namely the "output" properties; the latter are initialized to default values instead. As with clone(), the new object can be provided in the optional argument CONTEXT (if CONTEXT is an object of the same class); otherwise a brand new object is used. Only properties recognized by CGI::Portable are set in this object; others are not touched.
take_context_output( CONTEXT[, LEAVE_SCALARS[, REPLACE_LISTS]] )
This method takes another CGI::Portable (or subclass) object as its CONTEXT argument and copies some of its properties to this object, potentially overwriting any versions already in this object. If CONTEXT is not a valid CGI::Portable (or subclass) object then this method returns without changing anything. The properties that get copied are the "output" properties that presumably need to work their way back to the user. In other words, this method copies everything that make_new_context() did not. This method will never copy any properties which are undefined scalars or empty lists, so a CONTEXT with no "output" properties set will not cause any changes. If any scalar output properties of CONTEXT are defined, they will overwrite any defined corresponding properties of this object by default; however, if the optional boolean argument LEAVE_SCALARS is true, then the scalar values are only copied if the ones in this object are not defined. If any list output properties of CONTEXT have elements, then they will be appended to any corresponding ones of this object by default, thereby preserving both (except with hash properties, where like hash keys will overwrite); however, if the optional boolean argument REPLACE_LISTS is true, then any existing list values are overwritten by any copied CONTEXT equivalents.
call_component( COMP_NAME )
This method can be used by one component to invoke another. For this to work, the called component needs to be a Perl 5 module with a method called main(). The argument COMP_NAME is a string containing the name of the module to be invoked. This method will first "require [COMP_NAME]" and then invoke its dispatch method with a "[COMP_NAME]->main()". These statements are wrapped in an "eval" block and if there was a compile or runtime failure then this method will log an error message like "can't use module '[COMP_NAME]': $@" and also set the output page to be an error screen using that. So regardless of whether the component worked or not, you can simply print the output page the same way. The call_component() method will pass a reference to the CGI::Portable object it is invoked from as an argument to the main() method of the called module. If you want the called component to get a different CGI::Portable object then you will need to create it in your caller using make_new_context() or new() or clone(). Anticipating that your component would fail because of it, this method will abort with an error screen prior to any "require" if there are errors already logged and unresolved. Any errors existing now were probably set by set_prefs(), meaning that the component would be missing its config data were it started up. This method will return 0 upon making an error screen; otherwise, it will return 1 if everything worked. Since this method calls add_no_error() upon making the error screen, you should pay attention to its return value if you want to make a custom screen instead (so you know when to).
METHODS FOR DEBUGGING
is_debug([ VALUE ])
This method is an accessor for the "is debug" boolean property of this object, which it returns. If VALUE is defined, this property is set to it. If this property is true then it indicates that the program is currently being debugged by the owner/maintainer; if it is false then the program is being run by a normal user. How or whether the program reacts to this fact is quite arbitrary. For example, it may just keep a separate set of usage logs or append "debug" messages to email or web pages it makes.
METHODS FOR SEARCH AND REPLACE
This method supplements the page_search_and_replace() method in CGI::Portable::Response with a more proprietary solution.
search_and_replace_url_path_tokens([ TOKEN ])
This method performs a specialized search-and-replace of this object's "page body" property. The nature of this search and replace allows you to to embed "url paths" in static portions of your application, such as data files, and then replace them with complete self-referencing urls that go to the application screen that each url path corresponds to. How it works is that your data files are formatted like '<a href="__url_path__=/pics/green">green pics</a>' or '<a href="__url_path__=../texts">texts page</a>' or '<a href="__url_path__=/jump&url=http://www.cpan.org">CPA/a' and the scalar argument TOKEN is equal to '__url_path__' (that is its default value also). This method will search for text like in the above formats, specifically the parts between the double-quotes, and substitute in self-referencing urls like '<a href="http://www.aardvark.net/it.pl/pics/green">green pics</a>' or '<a href="http://www.aardvark.net/it.pl/jump?url=http://www.cpan.org">CPA/a'. New urls are constructed in a similar fashion to what url_as_string() makes, and incorporates your existing url base, query string, and so on. Any query string you provide in the source text is added to the url query in the output. This specialized search and replace can not be done with page_search_and_replace() since that would only replace the '__url_path__' part and leave the rest. The regular expression that is searched for looks sort of like /"TOKEN=([^&^"]*)&?(.*?)"/.
METHODS FOR GLOBAL PREFERENCES
These methods are designed to be accessors for a few "special" preferences that are global in the sense that they are stored separately from normal preferences and they only have to be set once in a parent context to be available to all child contexts and the application components that use them. Each one has its own accessor method. The information stored here is of the generic variety that could be used all over the application, such as the name of the application instance or the maintainer's name and email address, which can be used with error messages or other places where the maintainer would be contacted.
default_application_title([ VALUE ])
This method is an accessor for the "app instance title" string property of this object, which it returns. If VALUE is defined, this property is set to it. This property can be used on about/error screens or email messages to indicate the title of this application instance. You can call url_base() or recall_url() to provide an accompanying url in the emails if you wish. This property defaults to "Untitled Application".
default_maintainer_name([ VALUE ])
This method is an accessor for the "maintainer name" string property of this object, which it returns. If VALUE is defined, this property is set to it. This property can be used on about/error screens or email messages to indicate the name of the maintainer for this application instance, should you need to credit them or know who to contact. This property defaults to "Webmaster".
default_maintainer_email_address([ VALUE ])
This method is an accessor for the "maintainer email" string property of this object, which it returns. If VALUE is defined, this property is set to it. This property can be used on about/error screens or email messages to indicate the email address of the maintainer for this application instance, should you need to contact them or should this application need to send them an email. This property defaults to "webmaster@localhost".
default_maintainer_email_screen_url_path([ VALUE ])
This method is an accessor for the "maintainer screen" string property of this object, which it returns. If VALUE is defined, this property is set to it. This property can be used on about/error pages as an "url path" that goes to the screen of your application giving information on how to contact the maintainer. This property defaults to undefined, which means there is no screen in your app for this purpose; calling code that wants to use this would probably substitute the literal email address instead.
default_smtp_host([ VALUE ])
This method is an accessor for the "smtp host" string property of this object, which it returns. If VALUE is defined, this property is set to it. This property can be used by your application as a default web domain or ip for the smtp server that it should use to send email with. This property defaults to "localhost".
default_smtp_timeout([ VALUE ])
This method is an accessor for the "smtp timeout" number property of this object, which it returns. If VALUE is defined, this property is set to it. This property can be used by your application when contacting an smtp server to say how many seconds it should wait before timing out. This property defaults to 30.
maintainer_email_html([ LABEL ])
This method will selectively make a hyperlink that can be used by your users to contact the maintainer of this application. If the "maintainer screen" property is defined then this method will make a hyperlink to that screen. Otherwise, it makes an "mailto" hyperlink using the "maintainer email" address.
METHODS FOR MISCELLANEOUS OBJECT SERVICES
get_misc_objects_ref()
This method returns a reference to this object's "misc objects" hash property. This hash stores references to any objects you want to pass between program components with services that are beyond the scope of this class, such as persistent database handles. This hash ref is static across all objects of this class that are derived from one another.
replace_misc_objects( HASH_REF )
This method lets this object have a "misc objects" property in common with another object that it doesn't already. If the argument HASH_REF is a hash ref, then this property is set to it.
separate_misc_objects()
This method lets this object stop having a "misc objects" property in common with another, by replacing that property with a new empty hash ref.
AUTHOR
Copyright (c) 1999-2003, Darren R. Duncan. All rights reserved. This module is free software; you can redistribute it and/or modify it under the same terms as Perl itself. However, I do request that this copyright information and credits remain attached to the file. If you modify this module and redistribute a changed version then please attach a note listing the modifications. This module is available "as-is" and the author can not be held accountable for any problems resulting from its use.
I am always interested in knowing how my work helps others, so if you put this module to use in any of your own products or services then I would appreciate (but not require) it if you send me the website url for said product or service, so I know who you are. Also, if you make non-proprietary changes to the module because it doesn't work the way you need, and you are willing to make these freely available, then please send me a copy so that I can roll desirable changes into the main release.
Address comments, suggestions, and bug reports to perl@DarrenDuncan.net.
SEE ALSO
perl(1), CGI::Portable::*, mod_perl, Apache, Demo*, HTML::FormTemplate, CGI, CGI::Screen, CGI::MxScreen, CGI::Application, CGI::BuildPage, CGI::Response, HTML::Mason.
1 POD Error
The following errors were encountered while parsing the POD:
- Around line 837:
Deleting unknown formatting code N<>
Deleting unknown formatting code N<>