Don't try to parse HTML with regular expressions. You will always almost find a partial solution, but it will break in more and more esoteric ways.

Instead do something like this:

use HTML::TokeParser::Simple;

# time passes...

my $parser = HTML::TokeParser::Simple->new( string => $original_html )
+;
while (my $token = $parser->get_token) {
  $token->rewrite_tag;
  if ($token->is_text) {
    my $t = $token->as_is;
    $t =~ s/"/&quote;/g;  # More generically you could URI::Escape
    print $t;
  }
  else {
    print $token->as_is;
  }
}
[download]

use HTML::TokeParser::Simple;

# time passes...

my $parser = HTML::TokeParser::Simple->new( string => $original_html )
+;
while (my $token = $parser->get_token) {
  $token->rewrite_tag;
  if ($token->is_text) {
    my $t = $token->as_is;
    $t =~ s/"/&quote;/g;
    print $t;
  }
  elsif ($token->is_start_tag("script")) {
    # Embedded JavaScript, do not mess up!
    print $token->as_is;
    while ($token = $parser->get_token) {
      print $token->as_is;
      if ($token->is_tag("script") and not $token->is_start_tag) {
        last; # done with JavaScript
      }
    }
  }
  else {
    print $token->as_is;
  }
}
[download]

I had to do the same thing a couple of days ago. Here is my HTML::Parser solution (to complement tilly's HTML::TokeParser version above). It encodes any special characters found in the text portion of an HTML doc.

use HTML::Parser;
use HTML::Entities;

my $html = '<div align="center">Your "HTML" page goes here</div>';
my $enc = '';

my $p = HTML::Parser->new(
  unbroken_text => 1,
  default_h => [ sub { $enc .= join('', @_) }, "text" ],
  text_h => [ sub { $enc .= HTML::Entities::encode_entities($_[0]) }, 
+"text" ],
);
$p->parse($html);
print $encoded;
[download]

Handling JavaScript is left as an exercise for the reader ;)

- Cees