I'm drowning in coderefs in my HTML element module

Lady_Aleena has asked for the wisdom of the Perl Monks concerning the following question:

Please forgive me my brothers and sisters for this fourth node about my HTML::Element module. (1st, 2nd, 3rd with this tangentially related with a GitHub gist to hold the entire module so I do not copy it over and over again here.)

To the point, I am using too many (nested) coderefs. I have found going too deep with coderefs can have some nasty effects, such as losing variable values if one goes too deep. I try to have all my data munged before I even begin using my HTML element functions, however, if my variables become undefined at certain depths, it would be difficult to keep from mixing my logic from my display.

Initially, I had only two functions I use often with coderefs, however, there are six (or more) new tags which coderefs seem to be the only way to go. I am to the point of thinking about adding some arbitrary rules, but I am trying to avoid it.

Here is a sample function. (body can be replaced with div, article, section, aside, nav, header, footer.)

sub body {
  my ($tab,$code,$opt) = @_;
  my $tag = 'body';
  my $open = open_tag($tag,$opt,[@ics,@java]);

  line($tab,"<$open>");
  &$code;
  line($tab,"</$tag>");
}
[download]

The possible (and abbreviated) usage could be...

body( sub {
  header( sub {
    ... some header code for the body ...
  });
  ... some random code in the body ...
  article( sub {
    header( sub {
      ... some header code for the section...
    });
    ... some random code in the article ...
    section( sub {
      header( sub {
        ... some header code for the section ...
      });
      ... some random code in the section ...
      aside( sub {
        header( sub {
          ... some header code for the aside ...
        });
        ... some random code in the aside...
        footer( sub {
          ... some footer code for the aside ...
        });
      });
      ... some random code in the section...
      footer( sub {
        ... some footer code for the section...
      });
    });
    ... some random code in the article...
    section( sub {
      header( sub {
        ... some heading code for this section...
      });
      ... some random code for this section...
      footer( sub {
        ... some footer code for this section...
      });
    });
    
    aside( sub {
      header( sub {
        ... some header code for this aside ...
      });
      ... some random code for this aside ...
      footer( sub {
        ... some footer code for this aside ...
      });
    });
    
    footer( sub {
      ... some footer code for the article ...
    });
  });
  
  footer( sub {
    ... some footer code for the body ...
  });
});
[download]

Some of those arbitrary rules I was thinking about could lead me to rewrite all but header and footer to look something like...

sub body {
  my ($tab,$opt) = @_;
  my $tag = 'body';
  my $open = open_tag($tag,$opt,[@ics,@java]);

  line($tab,"<$open>");
  header($tab + 1, @{$opt->{header}}) if $opt->{header};
  &{$opt->{contents}};
  footer($tab + 1, @{$opt->{footer}}) if $opt->{footter};
  line($tab,"</$tag>");
}
[download]

The above could make the code look something like the following, however there are still too many coderefs.

body(
  header => sub {
    ... header code for body ...
  },
  footer => sub {
    address( sub {
      ... adress code for the footer of the body ...
    });
    ... footer code for body ...
  },
  contents => sub {
    ... random code for body ...
    article(
      header => sub {
        ... header code for article ...
      },
      footer => sub {
        ... footer code for article ...
      },
      contents => sub {
        ... random code for article ...
        section(
          header => sub {
            ... header code for section ...
          },
          footer => sub {
            ... footer code for section ...
          },
          contents => sub {
            ... random code for section ...
            aside(
              header => sub {
                ... header code for aside ...
              },
              footer => sub {
                ... footer code for aside ...
              },
              contents => sub {
                ... random code for aside ...
              },
            );
          },
        );
        ... random code for article ...
        section(
          header => sub {
            ... header code for section ...
          },
          footer => sub {
            ... footer code for section ...
          },
          contents => sub {
            ... random code for section ...
          },
        );
        ... random code for article ...
        aside(
          header => sub {
            ... header code for aside ...
          },
          footer => sub {
            ... footer code for aside ...
          },
          contents => sub {
            ... random code for aside ...
          },
        );
        ... random code for article ...
      },
    );
  },
);
[download]

As you can see, way too many coderefs to get one lost inside a script. I am a little lost right now after looking at it too long. I do not think any of my pages would ever get as complex as shown above, however, I am hoping to one day release this to CPAN for others to possibly use.

Have a cookie and a very nice day!

Lady Aleena

Comment on I'm drowning in coderefs in my HTML element module Select or Download Code

Replies are listed 'Best First'.
Re: I'm drowning in coderefs in my HTML element module by kcott (Archbishop) on Apr 30, 2013 at 10:04 UTC
G'day Lady Aleena, My first thought on this was to have a single, recursive routine that received the tag name as an argument. I had a very brief look at all the threads that had gone before but I don't have time to read them in detail. Consider the following a technique, rather than a solution, which you might be able to integrate into what you have already. Here's `pm_la_html.pl`: #!/usr/bin/env perl use strict; use warnings; my $indent = ' ' x 4; my $web_page = [ [ html => { attr => { lang => 'en' }, content => [ [ head => { content => [ [ title => { content => 'Some Title', }, ], ], }, ], [ body => { attr => { id => 'top' }, content => [ [ header => { }, ], [ article => { }, ], [ footer => { }, ], ], }, ], ], }, ], ]; markup($web_page); sub markup { my ($element_array, $depth) = @_; $depth \|\|= 0; for my $element (@$element_array) { my ($tag, $data) = @$element; print $indent x $depth, "<$tag"; if (exists $data->{attr}) { for my $key (keys %{$data->{attr}}) { print " $key=\"$data->{attr}{$key}\""; } } if (exists $data->{content}) { print ">\n"; if (ref $data->{content}) { markup($data->{content}, $depth + 1); } else { print $indent x ($depth + 1), "$data->{content}\n"; } print $indent x $depth, "</$tag>\n"; } else { print " />\n"; } } return; } [download] Here's the output: `$ pm_la_html.pl <html lang="en"> <head> <title> Some Title </title> </head> <body id="top"> <header /> <article /> <footer /> </body> </html>` [download] -- Ken	[reply] [d/l] [select]
Re: I'm drowning in coderefs in my HTML element module by Arunbear (Prior) on Apr 30, 2013 at 17:20 UTC
Template::Declare also uses anon subs to create HTML, so you could look at its source to see how it does this. Having said that, Template::Declare seems usage-wise much less verbose and feature-wise much more powerful than your module, so it's worth considering if it can fit your needs.	[reply]
Re: I'm drowning in coderefs in my HTML element module by hdb (Monsignor) on Apr 30, 2013 at 08:30 UTC
Your formatting standards are probably not optimal for this situation. I have taken the liberty to reformat it somewhat. Also I would propose to change the order always to header - content - footer which makes it a bit more readible as well (unless this has an impact on the output?). You can also avoid deep nesting by naming a few of the code refs and define them separately. Here is an illustration of what I mean, I hope this helps. # non-standard formatting ############################################ +############################# # always put footer at the end (assuming order does not matter?) body( header => sub { hgroup( sub { }); }, contents => sub { article( header => sub { }, contents => sub { section( header = +> sub { }, contents = +> sub { aside( header => sub { }, + contents => sub { }, + footer => sub { }, ); }, footer = +> sub { }, ); section( header = +> sub { }, contents = +> sub { }, footer = +> sub { }, ); aside( header = +> sub { }, contents = +> sub { }, footer = +> sub { }, ); }, footer => sub { }, ); }, footer => sub { address( sub { }); }, ); # alternative ######################################################## +############################# # remaining super-structure body( header => sub { hgroup( sub { }); }, contents => sub { article( header => sub { }, contents => \&contents1, footer => sub { }, ); }, footer => sub { address( sub { }); }, ); # delegated content sub contents1 { section( header => sub { }, contents => sub { aside( header => sub { }, contents => sub { }, footer => sub { }, + ); }, footer => sub { }, ); section( header => sub { }, contents => sub { }, footer => sub { }, ); aside( header => sub { }, contents => sub { }, footer => sub { }, ); } [download]	[reply] [d/l]