=head1 NAME

Maypole::Manual::Model - Maypole Model Classes


Maypole's model classes provide an interface to your data store.
In principle Maypole can connect to pretty much any data source,
but the default model is based on the popular L<Class::DBI> object
interface that uses the near-universal L<DBI> Perl interface to databases.

=head2 Maypole::Model::CDBI

Maypole model classes are pretty evil. The very first thing a Maypole
model class will do in a Maypole application is to cause a load of
table-based classes to come into being, and then assimilate them.

What I mean by this is that when you set up a Maypole application, in
your driver class, you'll say something like this:


C<setup> is a Maypole method, and it hands its parameter to the model
class. In our case, the argument is a DBI connect string, because
that's what C<Maypole::Model::CDBI>, the C<Class::DBI>-based model
expects. C<Maypole::Model::CDBI> has a method called C<setup_database>
that creates all the C<Class::DBI> table classes after connecting to the
database with that connect string. It does this by using
C<Class::DBI::Loader>, a utility module which asks a database
about its schema and sets up classes such as C<BeerDB::Beer> to inherit
from C<Class::DBI>.

Now it gets interesting. The names of these classes are stashed away in
the application's configuration, and then Maypole forcibly has these
classes inherit from the model class. Our C<BeerDB::Beer> now inherits
both from C<Class::DBI> and C<Maypole::Model::CDBI>.

This is useful for two reasons. The first reason is that
C<Maypole::Model::CDBI> is stuffed full of C<Class::DBI> goodies that
make writing Maypole applications a lot easier:

    package Maypole::Model::CDBI;
    use base qw(Maypole::Model::Base Class::DBI);
    use Maypole::Model::CDBI::AsForm;
    use Class::DBI::FromCGI;  # probabyly broken . 
    use Class::DBI::Loader;
    use Class::DBI::AbstractSearch;
    use Class::DBI::Plugin::RetrieveAll;
    use Class::DBI::Pager;

We'll meet most of these goodies in the
L<Standard Templates and Actions|Maypole::Manual::StandardTemplates>
chapter, where we explain how C<Maypole::Model::CDBI> works.

The second reason why we want our table classes to inherit from
C<Maypole::Model::CDBI> is because it provides a useful set of 
default actions. So what's an action, and why are they useful?

=head2 Maypole::Model::CDBI::Plain

The 'Plain' maypole Model : C<Maypole::Model::CDBI> allows you

    package Foo;
    use 'Maypole::Application';

    Foo->setup([qw/ Foo::SomeTable Foo::Other::Table /]);

    # untaint columns and provide custom actions for each class

    Foo::SomeTable->untaint_columns(email => ['email'], printable => [qw/name description/]);

    Foo::Other::Table->untaint_columns ( ... );

    sub Foo::SomeTable::SomeAction : Exported {

        . . .


=head2 Extending a model class with actions

Maypole operates primarily by turning URLs into method calls on a model
class. All that the model stage of Maypole's operation does, when it
comes down to it, is maintain a mapping of tables to classes, and
despatch a HTTP request to a method call on the relevant table class. This
means that if you request a URL such as


Maypole's first stage of operation is to turn that into
C<BeerDB::Brewery-E<gt>delete(20)>. Now, it's slightly more complex than
that. Firstly because it doesn't actually pass the parameter C<20>, but
it passes an object representing row 20 in the database, but we can
gloss over that for the second. No, the real issue is that Maypole does
not allow you to call just any method in the table class; that would be
somewhat insecure. 

Instead, Maypole makes a distinction between the kind of methods that
only the class itself and other Perl code can call, and the kind of
methods that anyone can call from a URL. This latter set of methods are
called B<exported> methods, and exporting is done by means of Perl
attributes. You define a method to be exported like so:

    sub drink :Exported {

This will allow the user to access C</beerdb/beer/drink> over the web.
An exported method accompanied with a template to render its output is
sometimes called an B<action>. 

Maypole model classes like C<Maypole::Model::CDBI> come with a
relatively handy set of actions which are all you need to set up a CRUD
(Create, Read, Update, Delete)
database front-end: viewing a row in a database, editing it, adding a
new one, deleting, and so on. The most important thing about Maypole,
though, is that it doesn't stop there. You can add your own.

For instance, in our beer database application, we could create a
C<BeerDB::Beer> package, and put some additional actions in there.

    package BeerDB::Beer;
    sub top_five :Exported {
        my ($class, $r) = @_;
        $r->objects([ ($r->retrieve_all_sorted_by("score"))[-5..-1] ]);

Our action is called as a class method with the Maypole request object.
It uses the C<Class::DBI::Plugin::RetrieveAll> module's
C<retrieve_all_sorted_by> mixin to get the top five scoring beers, and
puts these in the C<objects> slot of the request of object. Next, the
view class will come along and process the C<top_five> template with
these five beers.

We'll look more at how to put together actions in the
L<Standard Templates and Actions|Maypole::Manual::StandardTemplates>
chapter and our case studies.

=head2 Links

Next L<Maypole View Classes|Maypole::Manual::View>,
Previous L<Introduction to Maypole|Maypole::Manual::About>