NAME

Bric::HTMLTemplate - Writing HTML::Template scripts and templates


INTRODUCTION

This document describes how to use Bricolage's HTML::Template templating system. To get the most out of this document you'll need to have some familiarity with Bricolage templating using Mason - see Bric::Templates and Bric::AdvTemplates for details. I'll try to keep the overlap between the those documents and this one to a minimum. It also helps to have an idea of how HTML::Template works outside of Bricolage - for that you can refer to HTML::Template's documentation.


TEMPLATES IN BRICOLAGE

The Bricolage system uses templates to produce output for stories when they are published. Most likely you'll be creating templates to format your stories as HTML pages but you can also use HTML::Template to output WML, XML, email and more.

Templates are created in the same category tree as your stories and media. When a story is published the category tree is searched for templates starting in the primary category for the story. The search proceeds up the tree until a matching template is found.

Bricolage allows you to create two types of templates - element-specific templates or generic templates. Element-specific templates are assigned to a single element (ex. Article, Page, Pull Quote, etc.). Generic templates are assigned to the category.


SCRIPTS AND TEMPLATES

HTML::Template works by separating Perl code from HTML design. In Bricolage this results in two types of template files - .pl script files and .tmpl template files. The script files contain Perl code. The template files contain a mix of HTML tags and HTML::Template's <TMPL_*> tags.

This divide between programming and design allows for a division of labor between programmers and designers. First, the programmer can setup a set of elements and scripts (.pl files). Usually the programmer will also create some bare-bones example templates (.tmpl files). Next the designers can edit the template files to match the desired design.

As an additional benefit, if per-category design changes are required a designer can create template files in each category that will automatically be used by the existing scripts in the root category. Of course, the same is true of script files but it is much more common to tweak the design by category than the code.


CHOOSING A BURNER

Bricolage decides which burner module to use - Mason or HTML::Template - by looking at the burner setting for the top-level story element being published. To start using HTML::Template to publish a story type go to Admin -> Elements, find the story element and set its burner to HTML::Template.

When you're creating templates you'll also see a pull-down called burner. This determines whether you're creating a Mason ``story.mc'', an HTML::Template ``story.pl'' script or an HTML::Template ``story.tmpl'' template.


AN EXAMPLE STORY TYPE

We'll examine a simple example story type called ``Story''. Here's the element tree for ``Story'':

   Story
        - Deck             (textbox attribute)
        + Page             (repeatable element)
             - Paragraph   (repeatable textbox attribute)
             - Pull Quote  (repeatable textbox attribute)

The Story element has one attribute called Deck and can contain any number of Page elements. Pages are composed of Paragraph attributes and Pull Quote attributes, both of which can be repeated.

If this doesn't immediately make sense then you should probably go check out Bric::ElementAdmin before continuing - it's hard to write templates if you don't understand elements!


CHOOSING A STRATEGY

Bricolage is an exceedingly flexible system and the HTML::Template burner is no exception - there are a number of different ways you can write scripts and templates for the Story element tree. I'll start with what I think is the easiest to understand and proceed to more complicated approaches pointing out the advantages and drawbacks along the way.


STRATEGY 1: ONE SCRIPT, ONE TEMPLATE

For a simple element tree you can often get away with just a single pair of files - a script and a template for the top-level element. Here's an example script file that could be used to setup variables and loops for the example story above.

THE SCRIPT - story.pl

    # get our template
    my $template = $burner->new_template(autofill => 0);
    # setup story title
    $template->param(title => $story->get_title);
    # get deck and assign it to a var
    $template->param(deck => $element->get_data('deck'));
    # setup the page break variable
    $template->param(page_break => $burner->page_break);
    # loop through pages building up @page_loop
    my @page_loop;
    my $page_count = 1;
    while (my $page = $element->get_container('page', $page_count)) {
        # build per-page element loop
        my @element_loop;
        foreach my $e ($page->get_elements()) {
            # push on a row for this element
            push(@element_loop, { $e->get_name => $e->get_data });
        }
        # push element_loop and a page_count on this loop
        push(@page_loop, { 
                          element_loop => \@element_loop,
                          page_count   => $page_count++
                         });
    }
    # finish the page_loop
    $template->param(page_loop => \@page_loop);
    # call output and return the results
    return $template->output();

There's a lot going on in the script above so we'll take it step by step. The first thing the script does is get a new $template object:

    # get our template
    my $template = $burner->new_template(autofill => 0);

You may be wondering where $burner came from. Every script has access to three global variables: $burner, $story and $element. The $burner object is an instance of the Bric::Util::Burner::Template class. The $story and $element variables are the same as in the Mason system - check out the Bric::Templates manpage for details.

The new_template() call (like all the $burner method calls) is documented in the Bric::Util::Burner::Template manpage. I've turned off autofill since we're doing all the hard work ourselves here. With autofill on the script would be two lines long which wouldn't teach you much about writing HTML::Template scripts! More on autofill later.

So, now that we have a template object we'll start by setting up some variables:

    # setup story title
    $template->param(title => $story->get_title);
    # get deck and assign it to a var
    $template->param(deck => $element->get_data('deck'));
    # setup the page break variable
    $template->param(page_break => $burner->page_break);

The title variable assignment should be fairly self-explanatory - it gets the $story's title and makes it available to the template. Next the deck attribute is retrieved from the $element using the get_data() method. Since there can only be one deck attribute - it's not marked as repeatable in the element tree - it's safe to assign it to a single variable. Finally, a special variable is setup to paginate the story - $burner->page_break returns a value that can be inserted into the output to break pages.

The next step should look very familiar if you've ever setup a nested loop in HTML::Template. If you haven't then it probably looks frightening. I'll try to ease you in slow:

    # loop through pages building up @page_loop
    my @page_loop;
    my $page_count = 1;
    while (my $page = $element->get_container('page', $page_count)) {

These lines setup the variables we'll need to build the page_loop. We need to use a loop for pages since there can be more than one inside the story element. The $page_count variable is used to call get_container() and also to allow the template to setup ``next page'' and ``previous page'' links.

The get_container() call returns container elements of the 'page' type. What's a container element? Well, unfortunately Bricolage is a bit confused about what to call things internally - what the external system refers to simply as an element the guts refer to as containers. To make matters worse attributes are internally referred to as data elements. That said, calling elements ``container elements'' is nicely descriptive since only elements can contain other elements.

Now that the loop is setup it's time to extract the page data:

        # build per-page element loop
        my @element_loop;
        foreach my $e ($page->get_elements()) {
            # push on a row for this element
            push(@element_loop, { $e->get_name => $e->get_data });
        }

First the code creates a new array to hold the element variables from this page. Next we loop through all the elements in the page with the get_elements() call - these elements will be paragraphs and pull quotes. Each element gets turned into a single row in the element_loop containing a single variable with the same name as the element.

For example, let's say we have a page with three paragraphs and a pull quote. After this loop is finished @element_loop will look something like:

    @element_loop = (
        { "paragraph"  => "text of paragraph one..."   },
        { "pull quote" => "text of pull quote one..."  },
        { "paragraph"  => "text of paragraph two..."   },
        { "paragraph"  => "text of paragraph three..." },
    );

As you know from your knowledge of HTML::Template this is the structure for a TMPL_LOOP. Once we've got this structure we push it onto the outer page_loop along with the current $page_count. The $page_count is also incremented here so that the next page will get requested for the next iteration.

       # push element_loop and a page_count on this loop
       push(@page_loop, { 
                           element_loop => \@element_loop,
                           page_count   => $page_count++
                        });
    }

A completed @page_loop for a two-page story might look something like:

    @page_loop = (
        { 
            element_loop => [
                { "paragraph"  => "text of paragraph one..."   },
                { "pull quote" => "text of pull quote one..."  },
                { "paragraph"  => "text of paragraph two..."   },
                { "paragraph"  => "text of paragraph three..." },
            ],
            page_count => 1
        },
        { 
            element_loop => [
                { "paragraph"  => "text of paragraph one..."   },
                { "paragraph"  => "text of paragraph two..." },
            ],
            page_count => 2
        }
    );

Which, as you might know is just the array of hashes of arrays of hashes structure that HTML::Template expects for nested loops.

    # finish the page_loop
    $template->param(page_loop => \@page_loop);
    # call output and return the results
    return $template->output();

Finally, we send the @page_loop data to the template and return the results of running the template.

THE TEMPLATE - story.tmpl

The template for our script matches the variables and loops setup in the script. It adds a very small amount of HTML formatting just so you can see where formatting might be added:

    <tmpl_loop page_loop>
       <html>
       <head>
         <title><tmpl_var title></title>
       </head>
       <body>
         <tmpl_if __first__>
            <h1><tmpl_var title></h1>
            <b><tmpl_var deck></b>
         </tmpl_if>
         <tmpl_loop element_loop>
            <tmpl_if paragraph>
              <p><tmpl_var paragraph></p>
            </tmpl_if>
            <tmpl_if name="pull quote">
              <p><blockquote><i>
                <tmpl_var name="pull quote">
              </i></blockquote></p>
            </tmpl_if>
         </tmpl_loop>
         <tmpl_unless __first__>
         <a href=<tmpl_var expr="prev_page_link(page_count)">>Previous Page</a>
         </tmpl_unless>
         <tmpl_unless __last__>
         <a href=<tmpl_var expr="next_page_link(page_count)">>Next Page</a>
         </tmpl_unless>
       </body>
       </html>
       <tmpl_var page_break>
    </tmpl_loop>

Most of this should be pretty self-explanatory but I'll highlight some of the more interesting bits. First, the template makes use of HTML::Template's ``loop_context_vars'' option which is on by default in Bricolage. This allows the template to make decisions based on the automatic loop variables __first__ and __last__:

         <tmpl_if __first__>
            <h1><tmpl_var title></h1>
            <b><tmpl_var deck></b>
         </tmpl_if>

This snippet is used to put the title line and deck on the first page only. This mysterious section sets up the next and previous links:

   <tmpl_unless __first__>
      <a href=<tmpl_var expr="prev_page_link(page_count)">>Previous Page</a>
   </tmpl_unless>
   <tmpl_unless __last__>
      <a href=<tmpl_var expr="next_page_link(page_count)">>Next Page</a>
   </tmpl_unless>

The use of __first__ and __last__ should be obvious enough - the first page doesn't get a previous page link and the last page doesn't get a next page link. This section also makes use of some helper functions provided to make linking between pages easier. We could do this without them though - something like this would produce equivalent results:

         <tmpl_unless __first__>
            <tmpl_if expr="page_count == 2">
              <a href="index.html">Previous Page</a>
            <tmpl_else>
              <a href="index<tmpl_var expr="page_count - 2">.html">
                 Previous Page
              </a>
            </tmpl_if>
         </tmpl_unless>
         <tmpl_unless __last__>
            <a href="index<tmpl_var page_count>.html">Next Page</a>
         </tmpl_unless>

Although that would only work if your output channel was setup to output files with names like ``index.html'' and ``index1.html''. The next_page_link() and prev_page_link() functions will work with any output channel settings.

The final bit of mystery in this template is the use of the magic page_break variable:

   <tmpl_var page_break>

If you remember back in the script this was setup with a call to $burner->page_break. Inserting this value in your output will tell Bricolage to insert a page break. Also, Bricolage is smart enough not to output a trailing blank page so you don't have to worry about the spacing after page_break in the loop.

CONCLUSION

This first example has shown how a simple story type can be formatted using a single script and a single template. The script is responsible for setting up the variables and loops that the template uses to format the story.

Here's an analysis of this approach:

Advantages
Disadvantages


STRATEGY 2: NO SCRIPT, ONE TEMPLATE

As I hinted at above new_template()'s autofill option can do a lot of work for you. Combined with the default script creation you can often get away without creating any scripts at all.

THE DEFAULT SCRIPT

The default script is used if Bricolage needs to publish an element for which no script file (.pl) exists but for which there is a template file (.tmpl). It consists of:

   return $burner->new_template()->output;

Since no options are specified to new_template() the autofill option defaults to on. In autofill mode new_template() fills in variables and loops for your element tree automatically.

Several types of variables and loops are created by autofill:

THE SCRIPT - story.tmpl

The script for use with this strategy is almost exactly the same as for strategy 1 (sneaky, huh?). There are a couple changes. First, the autofill code translates spaces in element names into underscores to make them easier to use - I could have done that in my earlier script but the script was complicated enough to begin with! Also, the autofill code provides ``is_$name'' variables inside the element_loops to make testing for the type of the row more obvious and more fool-proof. In STRATEGY 1 a paragraph with the sole contents ``0'' wouldn't have been printed! The horror!

   <tmpl_loop page_loop>
       <html>
       <head>
         <title><tmpl_var title></title>
       </head>
       <body>
         <tmpl_if __first__>
            <h1><tmpl_var title></h1>
            <b><tmpl_var deck></b>
         </tmpl_if>
         <tmpl_loop element_loop>
            <tmpl_if is_paragraph>
              <p><tmpl_var paragraph></p>
            </tmpl_if>
            <tmpl_if is_pull_quote>
              <p><blockquote><i>
                <tmpl_var pull_quote>
              </i></blockquote></p>
            </tmpl_if>
         </tmpl_loop>
         <tmpl_unless __first__>
         <a href=<tmpl_var expr="prev_page_link(page_count)">>Previous Page</a>
         </tmpl_unless>
         <tmpl_unless __last__>
         <a href=<tmpl_var expr="next_page_link(page_count)">>Next Page</a>
         </tmpl_unless>
       </body>
       </html>
       <tmpl_var page_break>
    </tmpl_loop>

CONCLUSION

This example demonstrates the real power of Bricolage's HTML::Template system. Here's a breakdown of this strategy:

Advantages
Disadvantages


STRATEGY THREE: NO SCRIPTS, MANY TEMPLATES

Sometimes a little extra work can go a long way. If you're building an element that will be used as a sub-element in a number of trees then it pays to split out the functionality into separate pieces. Bricolage supports this by allowing you to create a script (.pl) and a template (.tmpl) for every element.

This strategy will deal with just templates, relying on autofill to setup vars and loops. The next strategy will deal with customizing the scripts for multiple elements.

TEMPLATE - story.tmpl

Here's a revised story.tmpl to makes a call to the page element script/template:

   <tmpl_loop page_loop>
       <html>
       <head>
         <title><tmpl_var title></title>
       </head>
       <body>
         <tmpl_if __first__>
            <h1><tmpl_var title></h1>
            <b><tmpl_var deck></b>
         </tmpl_if>
         <tmpl_var page>
         <tmpl_unless __first__>
         <a href=<tmpl_var expr="prev_page_link(page_count)">>Previous Page</a>
         </tmpl_unless>
         <tmpl_unless __last__>
         <a href=<tmpl_var expr="next_page_link(page_count)">>Next Page</a>
         </tmpl_unless>
       </body>
       </html>
       <tmpl_var page_break>
    </tmpl_loop>

Notice that instead of the inner element_loop there's a single TMPL_VAR called ``page''. This tells autofill to make a call to the element script for the page element - page.pl. Of course, as we saw earlier, if this script doesn't exist then the default script is used:

    return $burner->new_template()->output;

TEMPLATE - page.tmpl

Here's the page template that outputs the body of the page:

    <tmpl_loop element_loop>
        <tmpl_if is_paragraph>
          <p><tmpl_var paragraph></p>
        </tmpl_if>
        <tmpl_if is_pull_quote>
          <p><blockquote><i>
        <tmpl_var pull_quote>
          </i></blockquote></p>
        </tmpl_if>
    </tmpl_loop>

This should look pretty familiar - it's the exact same text that was in the original story.tmpl! Autofill sets up the same loops and variables whether you're in an original template or a sub-template.

One thing to note is that you can't just move the header and footer generating code into the page template. Since the __first__ and __last__ variables are only valid inside the loop in story.tmpl they can't be used in page.tmpl. This might be addressed in the future but until then see the next strategy for a solution.

CONCLUSION

This strategy is a good one when you have elements that will be shared between template trees. Here's a breakdown:

Advantages
Disadvantages


STRATEGY 4: SCRIPTS AND TEMPLATES

The Bricolage system is all about flexibility. In Strategy 1 you got an up-close look at a script that handles the entire template setup process. Fortunately you don't need to do all that work just to add a small enhancement. For an example, let's fix the problem I mentioned at the end of Strategy 3 - the header and footer for the Page element were stuck in story.tmpl by their reliance on __first__ and __last__.

TEMPLATE - story.tmpl

Here's the desired story.tmpl:

   <tmpl_loop page_loop>
       <html>
       <head>
         <title><tmpl_var title></title>
       </head>
       <body>
         <tmpl_var page>
       </body>
       </html>
       <tmpl_var page_break>
    </tmpl_loop>

TEMPLATE - page.tmpl

And the new Page template:

    <tmpl_if first>
        <h1><tmpl_var title></h1>
        <b><tmpl_var deck></b>
    </tmpl_if>
    <tmpl_loop element_loop>
        <tmpl_if is_paragraph>
          <p><tmpl_var paragraph></p>
        </tmpl_if>
        <tmpl_if is_pull_quote>
          <p><blockquote><i>
        <tmpl_var pull_quote>
          </i></blockquote></p>
        </tmpl_if>
    </tmpl_loop>
    <tmpl_unless first>
       <a href=<tmpl_var expr="prev_page_link(page_count)">>Previous Page</a>
    </tmpl_unless>
    <tmpl_unless last>
       <a href=<tmpl_var expr="next_page_link(page_count)">>Next Page</a>
    </tmpl_unless>

You'll notice that the element_loop is unchanged. The header and footer expressions are the same except that __first__ and __last__ are now just plain first and last. This was done to emphasize that we're not using HTML::Template's automatic loop variables here.

SCRIPT - story.pl

The problem here is simple - we've got some variables in the Story that need to be made available to the Page. Also, we'd like to do this without having to do all the work of Strategy 1. Here's the first half of the solution in story.pl:

    my $template = $burner->new_template();
    # collect pages in @pages
    my @pages = grep { $_->has_name('page') } $element->get_elements();
    # build @page_loop by calling run_script with page_count and
    # page_total params.
    my $page_count = 1;
    my @page_loop;
    foreach my $page (@pages) {
      push @page_loop, { page => $burner->run_script($page, 
                                                     $page_count, 
                                                     scalar @page_loop) };
      $page_count++;
    }
    # replace autofilled page_loop with new one
    $template->param(page_loop => \@page_loop);
    # return the output
    return $template->output();

Basically this script does the same thing that autofill does but only for a single loop - page_loop. Additionally, instead of calling run_script() with just the element parameter it also supplies two arguments - $page_count and $page_total (computed from @page_loop).

SCRIPT - page.pl

Now that we've setup story.pl to pass parameters to the Page element we'll need a script that does something with them.

   my ($page_count, $page_total) = @_;
   my $template = $burner->new_template();
   # setup params
   if ($page_count == 0) {
      $template->param(first => 1);
   }
   if ($page_count == $page_total) {
      $template->param(last => 1);
   }
   $template->param(page_count => $page_count);
   # return output
   return $template->output();

As you can see parameters are passed to scripts just as they are to Perl subroutines - through @_. The script uses these parameters to setup the template params it needs.

CONCLUSION

This Strategy uses the full set of Bricolage HTML::Template tools we've seen so far - scripts, templates, autofill and run_script().

Advantages
Disadvantages


STRATEGY 5: RELATED MEDIA

So far things have been kept pretty simple - our example story type contains only text. Now let's add the possibility of including images in our story. The new tree will look like:

   Story
        - Deck             (textbox attribute)
        + Page             (repeatable element)
             - Paragraph   (repeatable textbox attribute)
             - Pull Quote  (repeatable textbox attribute)
             + Image       (repeatable related media element)
                - Caption  (textbox attribute)

The Image element is of the type ``Related Media'' and has one non-repeatable attribute called ``Caption''. Since it's a related media element it also has the ability to point to a media object. In this case the template will assume that object being pointed to is an image.

TEMPLATE: story.tmpl

To keep things simple we'll start with the template used to format the story in Strategy 2 with a small addition:

   <tmpl_loop page_loop>
       <html>
       <head>
         <title><tmpl_var title></title>
       </head>
       <body>
         <tmpl_if __first__>
            <h1><tmpl_var title></h1>
            <b><tmpl_var deck></b>
         </tmpl_if>
         <tmpl_loop element_loop>
            <tmpl_if is_paragraph>
              <p><tmpl_var paragraph></p>
            </tmpl_if>
            <tmpl_if is_pull_quote>
              <p><blockquote><i>
                <tmpl_var pull_quote>
              </i></blockquote></p>
            </tmpl_if>
            <tmpl_if is_image>
              <p><tmpl_var image></p>
            </tmpl_if>
         </tmpl_loop>
         <tmpl_unless __first__>
         <a href=<tmpl_var expr="prev_page_link(page_count)">>Previous Page</a>
         </tmpl_unless>
         <tmpl_unless __last__>
         <a href=<tmpl_var expr="next_page_link(page_count)">>Next Page</a>
         </tmpl_unless>
       </body>
       </html>
       <tmpl_var page_break>
    </tmpl_loop>

The addition is another conditional inside the page's element_loop:

            <tmpl_if is_image>
              <p><tmpl_var image></p>
            </tmpl_if>

This will make a call out to the Image element's script when output.

SCRIPT: image.tmpl

For the script we'll use the autogenerated autofill script. Here's the template to format the image:

    <img src="<tmpl_var link>">
    <tmpl_if caption>
       <br><font size=-1><tmpl_var caption></font>
    </tmpl_if>

By now we know to expect the ``caption'' variable - this is the Caption attribute that can be defined for the Image element. The ``link'' variable is a new feature of autofill - it corresponds to the following code:

    my $media = $element->get_related_media;
    $template->param(link => $media->get_primary_uri()) if $media;

The same technique works for Related Stories elements. I'll omit a full example but the same ``link'' variable is provided and can be used in much the same way (albeit without the <img> tag).

CONCLUSION

This strategy enables the use of key Bricolage features - related media and related stories. Given that users tend to like multimedia and hyperlinks between stories I'll omit an Advantages and Disadvantages section - you'll probably have to use this Strategy whether you like it or not!


STRATEGY 6: CATEGORY SCRIPTS AND TEMPLATES

Just as elements can have scripts and templates, each category may have one script and one template associated with it called ``category.pl'' and ``category.tmpl''.

This is the HTML::Template equivalent of Mason's autohandlers although there are some differences in behavior. For one, HTML::Template's category scripts don't stack like Mason autohandlers - the search ends when a script or template is found and only one is ever run. A way to do optional stacking might be added in the future if users request it. Another difference is that category scripts can't be used to setup variables for the element scripts - every script runs in a separate environment.

For an example, let's imagine that we want to put a blue box around every page in our output. Instead of putting this HTML into our templates we'll do it in a category template.

THE TEMPLATE - category.tmpl

  <html>
  <head><tmpl_var title></head>
  <body>
  <table bgcolor=blue cellspacing=5 border=0><tr><td>
     <tmpl_var content>
  </td></tr></table>
  </body>
  </html>

Fairly simple, right? One thing to note is that do the wrapping we had to move the <head> setup with the title var into the category template. This is an unfortunate fact-of-life for HTML since the <head> section has to precede the <body> most category templates will have to include it. As a result having stories setup their own <head> section is hard (although not impossible - clever use of run_script() could get you there).

Aside from that the only new feature is the magic ``content'' variable. Every category.tmpl must include the content variable to indicate where the content will be inserted for each page.

THE SCRIPT - category.pl

As you probably can predict autofill will fill in all the variables in the above template. However, for reference here's the script:

   my $template = $burner->new_template();
   $template->param(title   => $story->get_title,
                    content => $burner->content);
   return $template->output();

One note - the $element variable is not initialized to the story's element. As a result autofill doesn't setup any of the normal vars and loops associated with the element tree for the story. This usually a good thing - if you start trying to output story elements inside your category templates then you're headed the wrong way. However, you can always get the element from $story and pass it to new_template manually:

   my $template = $burner->new_template(element => $story->get_element);

CONCLUSION

Advantages
Disadvantages


THE END

Certainly there are many more ways to use HTML::Template in Bricolage than I've covered in this document. Your main tools - scripts, templates, the Bric::Util::Burner::Template API and the various objects available in Bricolage - are now yours for the taking. Go forth and bend Bricolage whimpering to your task!