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

##</code><code>##

if ($#ARGV != 4) {
 print "usage: run batch file 'run' not this one\n";
 exit;
}

my $wordfile = $ARGV[0];
my $textfile=$ARGV[3];
my $OutPutFile=$ARGV[4];

##</code><code>##

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

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

##</code><code>##

open (IF1,"$wordfile")|| die "cannot open the file";
open (PF, "$textfile")|| die "cannot open the file";
open (OF,">$OutPutFile")|| die "cannot open the file";

##</code><code>##

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

##</code><code>##

my $List1Ref=ReadDataInHash (*IF1);

##</code><code>##

my %words=ReadDataInHash($wordfile);

##</code><code>##

sub ReadDataInHash()

##</code><code>##

{
    my $x = shift;
    my %list1=();
    while (my $line =<$x>) {
        chomp $line;
        my @arr=split /\s/,$line;
        for (my $i=0;$i<=$#arr ;$i++) {
            if ($i==0) {
                $list1{$arr[$i]}={};
            }
            else{
                ${%{$list1{$arr[0]}}}{$arr[$i]} = 1;
            }
        }
    }
    return {%list1};
}

##</code><code>##

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;
}

##</code><code>##

while (my $line=<PF>) {
    chomp($line);
    my @arrAbs=split (/\|/,$line);
    my $ID=$arrAbs[0];
    my $Title=$arrAbs[1];
    my $Abs=$arrAbs[2];
    @arrAbs=split (/\./,$Abs);
    print OF"$ID|";

    for (my $SentenceNumber=0;$SentenceNumber<=$#arrAbs ;$SentenceNumber++) {
        my $i=$SentenceNumber+1;
        print OF "<".$i.">";
        my $Sentence=$arrAbs[$SentenceNumber];
        my @arrAbsSen=split (' ',$Sentence);

        foreach my $word(@arrAbsSen) {

#to match terms in the list, stored in %{$List1Ref}.
            if (exists(${%{$List1Ref}}{uc($word)})) { 
                print OF "$word ";
            }
            
            else {
                foreach my $p (sort keys (%{$List1Ref})) {
                    if (exists(${%{${%{$List1Ref}}{$p}}}{uc($word)})) { 
                        print OF "mainterm:$p:matchedterm:$word ";
                        last; 
                    }
                }
            }
        }
    @arrAbsSen=();
    }
    print OF "\n";
    @arrAbs=();
}

##</code><code>##

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";
}

##</code><code>##

#!/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__

##</code><code>##

#!/usr/bin/perl

use strict;
use warnings;

die "Usage: $0 <words file> [<file(s)>]\n" if @ARGV < 1;

{
    my %words;
    my $file=shift;
    open my $fh, '<', $file or die "Can't open `$file': $!\n";
    while (<$fh>) {
        chomp;
        my ($first, @rest)=split;
        $words{$first}{$_}=1 for @rest;
    }
    
    sub matchword {
        my $word=shift;
        my $uword=uc $word;
        return $word if $words{$uword};
        $words{$_}{$uword} and
          return "mainterm:$_:matchedterm:$word"
          for sort keys %words;
        return; # nothing if nothing is found
    }
}

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

    for (0..$#sentences) {
        print '<', $_+1, '>',
          join ' ', map matchword($_), split ' ', $sentences[$_];
    }
    print "\n";
}

__END__

##</code><code>##

while (<>) {
    chomp;
    my ($id, undef, $abs)=split /\|/;
    print "$id|";
    my $i;
    print '<', ++$i, '>',
      join ' ', map matchword($_), split
      for split /\./, $abs;
    print "\n";
}