blah blah foo blah <em>foo</em> blah <a href="foo.html">foo</a> blah blah

I'd want the first and second instances of the word "foo" to match, and
be substituted for, but not the third or fourth (because they're in the
anchor). Notice that foo can be in another element, as with EM in the
example above.

My reason for this requirement is that the substitution I want to do is
to make an occurence of the sought-after string into an anchor,
replacing, for example, 'foo' with '<a href="foo.html">foo</a>'. The
problem is I want to do this in a situation where the program can be run
again and again with the same input text, so I don't want text that's
already been made an anchor to be modified again. (I.e., I don't want

<a href="<a href="foo.html">foo</a>.html"><a href="foo.html">foo</a></a>

the second time through.)

I hope I've made it clear what I'm after. Any help would be appreciated.

changewords.pl
--------------------
#!/usr/bin/perl -w
# changewords.pl - make substitutions in normal text of HTML files

sub usage { die "Usage: $0 <from> <to> <file> ...\n" }

my $from = shift or usage;
my $to = shift or usage;

# Build the HTML::Filter subclass to do the substituting.

package MyFilter;

use HTML::Entities qw(decode_entities encode_entities);

sub text {
 my $self = shift;
 my $text = decode_entities($_[0]);
 $text =~ s/\Q$from/$to/go; #most important line
 $self->SUPER::text(encode_entities($text));

package main;

 MyFilter->new->parse_file($_);

I've needed to practice this, so here you go (needs some error
checking):

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

package FilterTextNoAnchors;
use base qw/HTML::Filter/;

sub new {
  my $invocant = shift;
  my $filter = shift;    # sub ref to filter $_
  my $class = ref $invocant || $invocant;

  $self->{filter} = $filter;
  return $self;

__DATA__
blah blah foo blah <em>foo</em> blah <a href="foo.html">foo</a> blah blah
__END__

Hope that helps you as much as it helped me to write it!