Ah, you specified that words had to be removed, not tokens that could be part of a word. For the token 'bin' which should be removed:

1. foobin
2. binary
3. bin1

If all of those should be deleted then you can change that pattern from:

my $pattern = '\b(?:' . join('|', @tokens) . ')\b';
[download]

To:

my $pattern = join('|', @tokens);
[download]

If you only want to match words that start with 'bin', and are followed only by non-alpha characters, then this:

my $pattern = '^(?:' . join('|', @tokens) . ')[^a-zA-Z]*$';
[download]

A revised copy that handles the deletion of tokens with a purely line based input:

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

if (@ARGV != 3) {
    print "Usage: $0 <pattern file> <input file> <output file>\n";
    exit;
}

my ($pattern_filename, $source_filename, $dest_filename) = @ARGV;

open my $pattern_fh, '<', $pattern_filename or die "Failed to open $pa
+ttern_filename: $!";
my @tokens = ();
while (my $line = <$pattern_fh>) {
    chomp $line;
    push @tokens, $line;
}
my $pattern = '^(?:' . join('|', @tokens) . ')[^a-zA-Z]*$';
print "Search pattern: $pattern\n";

open my $infile, "<",  $source_filename or die "Failed to open $source
+_filename: $!";
open my $outfile,">>", $dest_filename   or die "Failed to open $dest_f
+ilename: $!";

while(my $line = <$infile>) {
    print "input : $line";
    if ($line =~ /$pattern/) {
        next;
    }
    print "output: $line";
    print $outfile $line;
}
close($infile);
close($outfile);
[download]