Note that the output is utf8 because of the  .

#!/usr/local/bin/perl

use strict;
use warnings;
use HTML::Entities;
use HTML::TokeParser::Simple;

my $p = HTML::TokeParser::Simple->new(*DATA);

my $span_start;
while (my $t = $p->get_token){
  $span_start++, last if (
    $t->is_start_tag(q{span})
      and
    $t->get_attr(q{class})
      and
    $t->get_attr(q{class}) eq q{classification_indent}
  );
}
die qq{start span not found\n} unless $span_start;

my $span_end;
while (my $t = $p->get_token){
  $span_end++, last if $t->is_end_tag(q{span});
}
die qq{close span not found\n} unless $span_end;

my ($div_end, $encoded);
while (my $t = $p->get_token){
  $div_end++, last if $t->is_end_tag(q{div});
  $encoded .= $t->as_is if $t->is_text;
}
die qq{close div not found\n} unless $div_end;
die qq{no text found\n} unless $encoded;

my $decoded = decode_entities $encoded; # returns utf8

for ($decoded){
  s/^\W+//;
  s/\W+$//;
}
printf qq{*%s*\n}, $decoded;



__DATA__
     <span class="classification_indent">Advertiser:</span>&nbsp; 
        Epicor AS
      

</div> </td>
[download]

*Epicor AS*
[download]