#!/usr/bin/perl -w

# keywords.pl - by dual
#
# Generates a reverse keyword index
###################################


use strict;


# Declare
my $file = shift;
my @array;
my @unique;
my %seen;


# Open working files
open FILE, "<$file" or die "Can't open file: $!";
open INDEX, ">>/some/dir/keyword.index" or die "Can't open index: $!";


# Do work
while (<FILE>) {
	chomp;
	s/\s+/ /g;
	if ($_ !~ /^\s+$/) {
		my @temp = split(/ /, $_);
		foreach my $word (@temp) {
			if ($word !~ /
					\ba\b|
					\ball\b|
					\balso\b|
					\ban\b|
					\band\b|
					\bany\b|
					\bare\b|
					\bas\b|
					\bbe\b|
					\bbeen\b|
					\bbut\b|
					\bby\b|
					\bcan\b|
					\bcan't\b|
					\bdo\b|
					\bdoes\b|
					\bdoesn't\b|
					\bdon't\b|
					\belse\b|
					\beven\b|
					\bevery\b|
					\bfor\b|
					\bget\b|
					\bhad\b|
					\bhas\b|
					\bhave\b|
					\bhere\b|
					\bi\b|
					\bif\b|
					\bin\b|
					\bis\b|
					\bisn't\b|
					\bit\b|
					\blike\b|
					\bmany\b|
					\bmay\b|
					\bme\b|
					\bmore\b|
					\bmuch\b|
					\bmy\b|
					\bnot\b|
					\bof\b|
					\boff\b|
					\bon\b|
					\bonly\b|
					\bor\b|
					\bput\b|
					\bso\b|
					\bsome\b|
					\bthan\b|
					\bthat\b|
					\bthe\b|
					\btheir\b|
					\bthem\b|
					\bthen\b|
					\bthere\b|
					\bthey\b|
					\bthis\b|
					\bthose\b|
					\bto\b|
					\bvery\b|
					\bwas\b|
					\bwe\b|
					\bwhat\b|
					\bwhen\b|
					\bwhich\b|
					\bwho\b|
					\bwill\b|
					\bwith\b|
					\bwon't\b|
					\byou\b|
					\byour\b|
					\byou're\b
				/ix) {
				next if $seen{ $word }++;
				push @unique, $word;
			}
		}
	}
}


# Output
print INDEX "$file=>@unique\n";


# Clean up
close FILE;
close INDEX;