Template/Tutorial/Web.pod

#============================================================= -*-perl-*-
#
# Template::Tutorial::Web
#
# DESCRIPTION
#   Tutorial on generating web content with the Template Toolkit
#
# AUTHOR
#   Andy Wardley  <abw@wardley.org>
#
# COPYRIGHT
#   Copyright (C) 1996-2008 Andy Wardley.  All Rights Reserved.
#
#   This module is free software; you can redistribute it and/or
#   modify it under the same terms as Perl itself.
#
#========================================================================

=head1 NAME

Template::Tutorial::Web - Generating Web Content Using the Template Toolkit

=head1 Overview

This tutorial document provides a introduction to the Template Toolkit and
demonstrates some of the typical ways it may be used for generating web
content. It covers the generation of static pages from templates using the
L<tpage|Template::Tools::tpage> and L<ttree|Template::Tools::ttree> scripts
and then goes on to show dynamic content generation using CGI scripts and
Apache/mod_perl handlers.

Various features of the Template Toolkit are introduced and described briefly
and explained by use of example. For further information, see L<Template>,
L<Template::Manual> and the various sections within it. e.g

    perldoc Template                    # Template.pm module usage
    perldoc Template::Manual            # index to manual
    perldoc Template::Manual::Config    # e.g. configuration options

The documentation is also available in HTML format to read online, or download
from the Template Toolkit web site:

    http://template-toolkit.org/docs/

=head1 Introduction

The Template Toolkit is a set of Perl modules which collectively
implement a template processing system.

A template is a text document with special markup tags embedded in it.
By default, the Template Toolkit uses 'C<[%>' and 'C<%]>' to denote
the start and end of a tag.  Here's an example:

    [% INCLUDE header %]

    People of [% planet %], your attention please.

    This is [% captain %] of the
    Galactic Hyperspace Planning Council.

    As you will no doubt be aware, the plans
    for development of the outlying regions
    of the Galaxy require the building of a
    hyperspatial express route through your
    star system, and regrettably your planet
    is one of those scheduled for destruction.

    The process will take slightly less than
    [% time %].

    Thank you.

    [% INCLUDE footer %]

Tags can contain simple I<variables> (like C<planet> and C<captain>) and more
complex I<directives> that start with an upper case keyword (like C<INCLUDE>).
A directive is an instruction that tells the template processor to perform
some action, like processing another template (C<header> and C<footer> in this
example) and inserting the output into the current template. In fact, the
simple variables we mentioned are actually C<GET> directives, but the C<GET>
keyword is optional.

    People of [% planet %], your attention please.      # short form
    People of [% GET planet %], your attention please.  # long form

Other directives include C<SET> to set a variable value (the C<SET> keyword is
also optional), C<FOREACH> to iterate through a list of values, and C<IF>,
C<UNLESS>, C<ELSIF> and C<ELSE> to declare conditional blocks.

The Template Toolkit processes all I<text> files equally, regardless of what
kind of content they contain.  So you can use TT to generate HTML, XML, CSS,
Javascript, Perl, RTF, LaTeX, or any other text-based format.  In this tutorial,
however, we'll be concentrating on generating HTML for web pages.

=head1 Generating Static Web Content

Here's an example of a template used to generate an HTML document.

    [%  INCLUDE header
          title = 'This is an HTML example';

        pages = [
          { url   = 'http://foo.org'
            title = 'The Foo Organisation'
          }
          { url   = 'http://bar.org'
            title = 'The Bar Organisation'
          }
        ]
    %]
       <h1>Some Interesting Links</h1>
       <ul>
    [%  FOREACH page IN pages %]
         <li><a href="[% page.url %]">[% page.title %]</a>
    [%  END %]
       </ul>

    [% INCLUDE footer %]

This example shows how the C<INCLUDE> directive is used to load and process
separate 'C<header>' and 'C<footer>' template files, including the output in
the current document.  These files might look something like this:

header:

    <html>
      <head>
        <title>[% title %]</title>
      </head>
      <body>

footer:

        <div class="copyright">
          &copy; Copyright 2007 Arthur Dent
        </div>
      </body>
    </html>

The example also uses the C<FOREACH> directive to iterate through the
'C<pages>' list to build a table of links. In this example, we have defined
this list within the template to contain a number of hash references, each
containing a 'C<url>' and 'C<title>' member. The C<FOREACH> directive iterates
through the list, aliasing 'C<page>' to each item (in this case, hash array
references). The C<[% page.url %]> and C<[% page.title %]> directives then
access the individual values in the hash ararys and insert them into the
document.

=head2 Using tpage

Having created a template file we can now process it to generate some real
output. The quickest and easiest way to do this is to use the
L<tpage|Template::Tools::tpage> script. This is provided as part of the
Template Toolkit and should be installed in your usual Perl bin directory.

Assuming you saved your template file as F<example.html>, you would run
the command:

    $ tpage example.html

This will process the template file, sending the output to C<STDOUT> (i.e.
whizzing past you on the screen). You may want to redirect the output to a
file but be careful not to specify the same name as the template file, or
you'll overwrite it. You may want to use one prefix for your templates (e.g.
'C<.tt>') and another (e.g. 'C<.html>') for the output files.

    $ tpage example.tt > example.html

Or you can redirect the output to another directory. e.g.

    $ tpage templates/example.tt > html/example.html

The output generated would look like this:

    <html>
      <head>
        <title>This is an HTML example</title>
      </head>
      <body>
        <h1>Some Interesting Links</h1>
        <ul>
          <li><a href="http://foo.org">The Foo Organsiation</a>
          <li><a href="http://bar.org">The Bar Organsiation</a>
        </ul>
        <div class="copyright">
          &copy; Copyright 2007 Arthur Dent
        </div>
      </body>
    </html>

The F<header> and F<footer> template files have been included (assuming
you created them and they're in the current directory) and the link data
has been built into an HTML list.

=head2 Using ttree

The L<tpage|Template::Tools::tpage> script gives you a simple and easy way to
process a single template without having to write any Perl code. The
L<ttree:Template::Tools::ttree> script, also distributed as part of the
Template Toolkit, provides a more flexible way to process a number of template
documents in one go.

The first time you run the script, it will ask you if it should create a
configuration file (F<.ttreerc>) in your home directory. Answer C<y> to have
it create the file.

The L<ttree:Template::Tools::ttree> documentation describes how you can change
the location of this file and also explains the syntax and meaning of the
various options in the file. Comments are written to the sample configuration
file which should also help.

In brief, the configuration file describes the directories in which template
files are to be found (C<src>), where the corresponding output should be
written to (C<dest>), and any other directories (C<lib>) that may contain
template files that you plan to C<INCLUDE> into your source documents. You can
also specify processing options (such as C<verbose> and C<recurse>) and provide
regular expression to match files that you don't want to process (C<ignore>,
C<accept>)> or should be copied instead of being processed as templates (C<copy>).

An example F<.ttreerc> file is shown here:

$HOME/.ttreerc:

    verbose
    recurse

    # this is where I keep other ttree config files
    cfg = ~/.ttree

    src  = ~/websrc/src
    lib  = ~/websrc/lib
    dest = ~/public_html/test

    ignore = \b(CVS|RCS)\b
    ignore = ^#

You can create many different configuration files and store them
in the directory specified in the C<cfg> option, shown above.  You then
add the C<-f filename> option to C<ttree> to have it read that file.

When you run the script, it compares all the files in the C<src> directory
(including those in sub-directories if the C<recurse> option is set), with
those in the C<dest> directory.  If the destination file doesn't exist or
has an earlier modification time than the corresponding source file, then
the source will be processed with the output written to the destination
file.  The C<-a> option forces all files to be processed, regardless of
modification times.

The script I<doesn't> process any of the files in the C<lib> directory, but it
does add it to the C<INCLUDE_PATH> for the template processor so that it can
locate these files via an C<INCLUDE>, C<PROCESS> or C<WRAPPER> directive.
Thus, the C<lib> directory is an excellent place to keep template elements
such as header, footers, etc., that aren't complete documents in their own
right.

You can also specify various Template Toolkit options from the configuration
file. Consult the L<ttree|Template::Tools::ttree> documentation and help
summary (C<ttree -h>) for full details. e.g.

$HOME/.ttreerc:

    pre_process = config
    interpolate
    post_chomp

The C<pre_process> option allows you to specify a template file which
should be processed before each file.  Unsurprisingly, there's also a
C<post_process> option to add a template after each file.  In the
fragment above, we have specified that the C<config> template should be
used as a prefix template.  We can create this file in the C<lib>
directory and use it to define some common variables, including those
web page links we defined earlier and might want to re-use in other
templates.  We could also include an HTML header, title, or menu bar
in this file which would then be prepended to each and every template
file, but for now we'll keep all that in a separate C<header> file.

$lib/config:

    [% root     = '~/abw'
       home     = "$root/index.html"
       images   = "$root/images"
       email    = 'abw@wardley.org'
       graphics = 1
       webpages = [
         { url => 'http://foo.org', title => 'The Foo Organsiation' }
         { url => 'http://bar.org', title => 'The Bar Organsiation' }
       ]
    %]

Assuming you've created or copied the C<header> and C<footer> files from the
earlier example into your C<lib> directory, you can now start to create
web pages like the following in your C<src> directory and process them
with C<ttree>.

$src/newpage.html:

    [% INCLUDE header
       title = 'Another Template Toolkit Test Page'
    %]

        <a href="[% home %]">Home</a>
        <a href="mailto:[% email %]">Email</a>

    [% IF graphics %]
        <img src="[% images %]/logo.gif" align=right width=60 height=40>
    [% END %]

    [% INCLUDE footer %]

Here we've shown how pre-defined variables can be used as flags to
enable certain feature (e.g. C<graphics>) and to specify common items
such as an email address and URL's for the home page, images directory
and so on.  This approach allows you to define these values once so
that they're consistent across all pages and can easily be changed to
new values.

When you run F<ttree>, you should see output similar to the following
(assuming you have the verbose flag set).

    ttree 2.9 (Template Toolkit version 2.20)

         Source: /home/abw/websrc/src
    Destination: /home/abw/public_html/test
   Include Path: [ /home/abw/websrc/lib ]
         Ignore: [ \b(CVS|RCS)\b, ^# ]
           Copy: [  ]
         Accept: [ * ]

    + newpage.html

The C<+> in front of the C<newpage.html> filename shows that the file was
processed, with the output being written to the destination directory. If you
run the same command again, you'll see the following line displayed instead
showing a C<-> and giving a reason why the file wasn't processed.

    - newpage.html                     (not modified)

It has detected a C<newpage.html> in the destination directory which is
more recent than that in the source directory and so hasn't bothered
to waste time re-processing it.  To force all files to be processed,
use the C<-a> option.  You can also specify one or more filenames as
command line arguments to C<ttree>:

    tpage newpage.html

This is what the destination page looks like.

$dest/newpage.html:

    <html>
      <head>
        <title>Another Template Toolkit Test Page</title>
      </head>
      <body>

        <a href="~/abw/index.html">Home</a>
        <a href="mailto:abw@wardley.org">Email me</a>
        <img src="~/abw/images/logo.gif" align=right width=60 height=40>

        <div class="copyright">
          &copy; Copyright 2007 Arthur Dent
        </div>
      </body>
    </html>

You can add as many documents as you like to the C<src> directory and
C<ttree> will apply the same process to them all.  In this way, it is
possible to build an entire tree of static content for a web site with
a single command.  The added benefit is that you can be assured of
consistency in links, header style, or whatever else you choose to
implement in terms of common templates elements or variables.

=head1 Dynamic Content Generation Via CGI Script

The L<Template> module provides a simple front-end to the Template Toolkit for
use in CGI scripts and Apache/mod_perl handlers. Simply C<use> the L<Template>
module, create an object instance with the L<new()> method and then call the
L<process()> method on the object, passing the name of the template file as a
parameter. The second parameter passed is a reference to a hash array of
variables that we want made available to the template:

    #!/usr/bin/perl
    use strict;
    use warnings;
    use Template;

    my $file = 'src/greeting.html';
    my $vars = {
       message  => "Hello World\n"
    };

    my $template = Template->new();

    $template->process($file, $vars)
        || die "Template process failed: ", $template->error(), "\n";

So that our scripts will work with the same template files as our earlier
examples, we'll can add some configuration options to the constructor to
tell it about our environment:

    my $template->new({
        # where to find template files
        INCLUDE_PATH => ['/home/abw/websrc/src', '/home/abw/websrc/lib'],
        # pre-process lib/config to define any extra values
        PRE_PROCESS  => 'config',
    });

Note that here we specify the C<config> file as a C<PRE_PROCESS> option.
This means that the templates we process can use the same global
variables defined earlier for our static pages.  We don't have to
replicate their definitions in this script.  However, we can supply
additional data and functionality specific to this script via the hash
of variables that we pass to the C<process()> method.

These entries in this hash may contain simple text or other values,
references to lists, others hashes, sub-routines or objects.  The Template
Toolkit will automatically apply the correct procedure to access these
different types when you use the variables in a template.

Here's a more detailed example to look over.  Amongst the different
template variables we define in C<$vars>, we create a reference to a
L<CGI> object and a C<get_user_projects()> sub-routine.

    #!/usr/bin/perl
    use strict;
    use warnings;
    use Template;
    use CGI;

    $| = 1;
    print "Content-type: text/html\n\n";

    my $file = 'userinfo.html';
    my $vars = {
        'version'  => 3.14,
        'days'     => [ qw( mon tue wed thu fri sat sun ) ],
        'worklist' => \&get_user_projects,
        'cgi'      => CGI->new(),
        'me'       => {
            'id'     => 'abw',
            'name'   => 'Andy Wardley',
        },
    };

    sub get_user_projects {
        my $user = shift;
        my @projects = ...   # do something to retrieve data
        return \@projects;
    }

    my $template = Template->new({
        INCLUDE_PATH => '/home/abw/websrc/src:/home/abw/websrc/lib',
        PRE_PROCESS  => 'config',
    });

    $template->process($file, $vars)
        || die $template->error();

Here's a sample template file that we might create to build the output
for this script.

$src/userinfo.html:

    [% INCLUDE header
       title = 'Template Toolkit CGI Test'
    %]

    <a href="mailto:[% email %]">Email [% me.name %]</a>

    <p>This is version [% version %]</p>

    <h3>Projects</h3>
    <ul>
    [% FOREACH project IN worklist(me.id) %]
       <li> <a href="[% project.url %]">[% project.name %]</a>
    [% END %]
    </ul>

    [% INCLUDE footer %]

This example shows how we've separated the Perl implementation (code) from the
presentation (HTML). This not only makes them easier to maintain in isolation,
but also allows the re-use of existing template elements such as headers and
footers, etc. By using template to create the output of your CGI scripts, you
can give them the same consistency as your static pages built via
L<ttree|Template::Tools::ttree> or other means.

Furthermore, we can modify our script so that it processes any one of a
number of different templates based on some condition.  A CGI script to
maintain a user database, for example, might process one template to
provide an empty form for new users, the same form with some default
values set for updating an existing user record, a third template for
listing all users in the system, and so on.  You can use any Perl
functionality you care to write to implement the logic of your
application and then choose one or other template to generate the
desired output for the application state.

=head1 Dynamic Content Generation Via Apache/Mod_Perl Handler

B<NOTE:> the L<Apache::Template> module is available from CPAN and provides a
simple and easy to use Apache/mod_perl interface to the Template Toolkit.
Although basic, it implements most, if not all of what is described below, and
it avoids the need to write your own handler. However, in many cases, you'll
want to write your own handler to customise processing for your own need, and
this section will show you how to get started.

The L<Template> module can be used from an Apache/mod_perl handler. Here's an
example of a typical Apache F<httpd.conf> file:

    PerlModule CGI;
    PerlModule Template
    PerlModule MyOrg::Apache::User

    PerlSetVar websrc_root   /home/abw/websrc

    <Location /user/bin>
        SetHandler     perl-script
        PerlHandler    MyOrg::Apache::User
    </Location>

This defines a location called C</user/bin> to which all requests will
be forwarded to the C<handler()> method of the C<MyOrg::Apache::User>
module.  That module might look something like this:

    package MyOrg::Apache::User;

    use strict;
    use vars qw( $VERSION );
    use Apache::Constants qw( :common );
    use Template qw( :template );
    use CGI;

    $VERSION = 1.59;

    sub handler {
        my $r = shift;

        my $websrc = $r->dir_config('websrc_root')
            or return fail($r, SERVER_ERROR,
                           "'websrc_root' not specified");

        my $template = Template->new({
            INCLUDE_PATH  => "$websrc/src/user:$websrc/lib",
            PRE_PROCESS   => 'config',
            OUTPUT        => $r,     # direct output to Apache request
        });

        my $params = {
            uri     => $r->uri,
            cgi     => CGI->new,
        };

        # use the path_info to determine which template file to process
        my $file = $r->path_info;
        $file =~ s[^/][];

        $r->content_type('text/html');
        $r->send_http_header;

        $template->process($file, $params)
            || return fail($r, SERVER_ERROR, $template->error());

        return OK;
    }

    sub fail {
        my ($r, $status, $message) = @_;
        $r->log_reason($message, $r->filename);
        return $status;
    }

The handler accepts the request and uses it to determine the C<websrc_root>
value from the config file.  This is then used to define an C<INCLUDE_PATH>
for a new L<Template> object.  The URI is extracted from the request and a
L<CGI> object is created.  These are both defined as template variables.

The name of the template file itself is taken from the C<PATH_INFO> element
of the request.  In this case, it would comprise the part of the URL
coming after C</user/bin>,  e.g for C</user/bin/edit>, the template file
would be C<edit> located in C<$websrc/src/user>.  The headers are sent
and the template file is processed.  All output is sent directly to the
C<print()> method of the Apache request object.

=head1 Using Plugins to Extend Functionality

As we've already shown, it is possible to bind Perl data and functions
to template variables when creating dynamic content via a CGI script
or Apache/mod_perl process.  The Template Toolkit also supports a
plugin interface which allows you define such additional data and/or
functionality in a separate module and then load and use it as
required with the C<USE> directive.

The main benefit to this approach is that you can load the extension into
any template document, even those that are processed "statically" by
C<tpage> or C<ttree>.  You I<don't> need to write a Perl wrapper to
explicitly load the module and make it available via the stash.

Let's demonstrate this principle using the C<DBI> plugin written by Simon
Matthews (available from CPAN). You can create this template in your C<src>
directory and process it using C<ttree> to see the results. Of course, this
example relies on the existence of the appropriate SQL database but you should
be able to adapt it to your own resources, or at least use it as a
demonstrative example of what's possible.

    [% INCLUDE header
         title = 'User Info'
    %]

    [% USE DBI('dbi:mSQL:mydbname') %]

    <table border=0 width="100%">
      <tr>
        <th>User ID</th>
        <th>Name</th>
        <th>Email</th>
      </tr>
    [% FOREACH user IN DBI.query('SELECT * FROM user ORDER BY id') %]
      <tr>
        <td>[% user.id %]</td>
        <td>[% user.name %]</td>
        <td>[% user.email %]</td>
      </tr>
    [% END %]
    </table>

    [% INCLUDE footer %]

A plugin is simply a Perl module in a known location and conforming to
a known standard such that the Template Toolkit can find and load it
automatically.  You can create your own plugin by inheriting from the
L<Template::Plugin> module.

Here's an example which defines some data items (C<foo> and C<people>)
and also an object method (C<bar>).  We'll call the plugin C<FooBar> for
want of a better name and create it in the C<MyOrg::Template::Plugin::FooBar>
package.  We've added a C<MyOrg> to the regular C<Template::Plugin::*> package
to avoid any conflict with existing plugins.

    package MyOrg::Template::Plugin::FooBar;
    use base 'Template::Plugin'
    our $VERSION = 1.23;

    sub new {
        my ($class, $context, @params) = @_;

        bless {
            _CONTEXT => $context,
            foo      => 25,
            people   => [ 'tom', 'dick', 'harry' ],
        }, $class;
    }

    sub bar {
        my ($self, @params) = @_;
        # ...do something...
        return $some_value;
    }

The plugin constructor C<new()> receives the class name as the first
parameter, as is usual in Perl, followed by a reference to something called a
L<Template::Context> object. You don't need to worry too much about this at
the moment, other than to know that it's the main processing object for the
Template Toolkit. It provides access to the functionality of the processor and
some plugins may need to communicate with it. We don't at this stage, but
we'll save the reference anyway in the C<_CONTEXT> member. The leading
underscore is a convention which indicates that this item is private and the
Template Toolkit won't attempt to access this member. The other members
defined, C<foo> and C<people> are regular data items which will be made
available to templates using this plugin. Following the context reference are
passed any additional parameters specified with the USE directive, such as the
data source parameter, C<dbi:mSQL:mydbname>, that we used in the earlier DBI
example.

If you don't or can't install it to the regular place for your Perl
modules (perhaps because you don't have the required privileges) then
you can set the PERL5LIB environment variable to specify another location.
If you're using C<ttree> then you can add the following line to your
configuration file instead.

$HOME/.ttreerc:

    perl5lib = /path/to/modules

One further configuration item must be added to inform the toolkit of
the new package name we have adopted for our plugins:

$HOME/.ttreerc:

    plugin_base = 'MyOrg::Template::Plugin'

If you're writing Perl code to control the L<Template> modules directly,
then this value can be passed as a configuration parameter when you
create the module.

    use Template;

    my $template = Template->new({
        PLUGIN_BASE => 'MyOrg::Template::Plugin'
    });

Now we can create a template which uses this plugin:

    [% INCLUDE header
       title = 'FooBar Plugin Test'
    %]

    [% USE FooBar %]

    Some values available from this plugin:
      [% FooBar.foo %] [% FooBar.bar %]

    The users defined in the 'people' list:
    [% FOREACH uid = FooBar.people %]
      * [% uid %]
    [% END %]

    [% INCLUDE footer %]

The C<foo>, C<bar>, and C<people> items of the FooBar plugin are
automatically resolved to the appropriate data items or method calls
on the underlying object.

Using this approach, it is possible to create application
functionality in a single module which can then be loaded and used on
demand in any template.  The simple interface between template
directives and plugin objects allows complex, dynamic content to be
built from a few simple template documents without knowing anything
about the underlying implementation.

=head1 AUTHOR

Andy Wardley E<lt>abw@wardley.orgE<gt> L<http://wardley.org/>

=head1 COPYRIGHT

Copyright (C) 1996-2007 Andy Wardley.  All Rights Reserved.

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

=cut

# Local Variables:
# mode: perl
# perl-indent-level: 4
# indent-tabs-mode: nil
# End:
#
# vim: expandtab shiftwidth=4: