Take a look at this HTML document, which HTML::TreeBuilder's dump() method has annotated each node of the HTML document with a unique identitifer: 
<html> @0
  <head> @0.0 (IMPLICIT)
  <body> @0.1 (IMPLICIT)
    <table supply="_aref::load_data"> @0.1.0
      <tr iterator="supply.Next"> @0.1.0.0
        <th> @0.1.0.0.0
          "name"
        <th> @0.1.0.0.1
          "age"
        <th> @0.1.0.0.2
          "weight"
        <td builder="_text::iterator.name"> @0.1.0.0.3
        <td builder="_text::iterator.age"> @0.1.0.0.4
        <td builder="_text::iterator.weight"> @0.1.0.0.5

[download]

It would be possible to tie each identifier to an anonymous subroutine: 


my %dynamic_html = (
 '@0.1.0.0.5' => sub { my $node = shift; $node->splice_content(0,1,'ne
+w text'}
}

[download]
for pure Perl and pure HTML templating (ie, absolutely no foreign elements in the HTML.

However, the problem would be that the mapping would have to change each time that the HTML changed the tree-numbering.

Back to Meditations