i'm assuming memory's not a problem here, since i've never seen html files larger than a few hundred kilobytes. this will slurp the file into a scalar, and match against the scalar. i have a feeling this is faster than matching against every line. i don't know if this will meet all your requirements, because i'm not sure what "then continue..." means.

#!/usr/bin/perl -w
use strict;

my $tmp = "/path/to/my.html";
my $pattern = qr/File Error/i;
my $html;
{
  local *INFILE;
  open( INFILE, $tmp ) or die "ack! $!";
  -f INFILE && sysread( INFILE, $html, -s INFILE );
}
die "$tmp: $pattern not found" unless $html =~ $pattern;
# then continue...
[download]

~Particle *accelerates*