#!/usr/bin/perl

use strict;
use warnings;

die "Usage: run batch file 'run' not this one\n"
  unless @ARGV==3;

my ($wordfile, $textfile, $outfile)=@ARGV;

open my $tfh, '<', $textfile or die "Can't open `$textfile': $!\n";
open my $ofh, '>', $outfile or die "Can't open `$outfile': $!\n";

my %words=ReadDataInHash($wordfile);

while (my $line=<$tfh>) {
    chomp $line;
    my ($id, $title, $abs)=split /\|/, $line;
    my @sentences=split /\./, $abs;
    print $ofh "$id|";

    for my $idx (0..$#sentences) {
        print $ofh '<', $idx+1, '>';
        for my $word (split ' ', $sentences[$idx]) {
            if ($words{uc $word}) {
                print $ofh "$word "
            }
            else {
                for (sort keys %words) {
                    if ($words{$_}{uc $word}) { 
                        print $ofh "mainterm:$_:matchedterm:$word ";
                        last; 
                    }
                }
            }
        }
    }
    print $ofh "\n";
}

sub ReadDataInHash {
    my $file=shift;
    open my $fh, '<', $file or die "Can't open `$file': $!\n";
    my %words;
    while (<$fh>) {
        chomp;
        my ($first, @rest)=split;
        $words{$first}{$_}=1 for @rest;
    }
    %words;
}

__END__