#!/usr/bin/perl

open(LIST, "erik_wordlist.txt") or die "Can't open word listing";
while (<LIST>) {
    ($word, $offset, $length, $amount) = split;
    $wordlist{$word} = "$offset $length $amount";
}
close(LIST);

# Open wordindex file for random access
open(INDEX, "erik_wordindex.txt") or die "Can't open word index file";
binmode(INDEX);


print "Enter query: ";
while (<STDIN>) {
    my($wordcount) = 0;
    last if m/^quit$/i or m/^q$/i;
    $_ = lc;
    @qwords = split;

    undef %queryset;
    foreach $qword (@qwords) {
	if (defined($wordlist{$qword})) {
	    my($offset, $length, $amount) = split(/\s/, $wordlist{$qword});
	    seek INDEX, $offset, 0;
	    read INDEX, $docnos, $length-1;
	    @docnos = split(/,/, $docnos);
	    $prevno = -1;
	    foreach $docno (@docnos) {
		if ($docno != $prevno) {
		    $queryset{$docno}++;
		}
		$prevno = $docno;  # weed out repetition
	    }
	}
	$wordcount++;
    }

    # Print documents that matched each key word
    $selection_index = 0;
    foreach $key (sort  {$a <=> $b} keys %queryset) {
	if ($queryset{$key} >= $wordcount) {
	    open(DOCS, "erik_docindex.txt");
	    for ($i = 0; $i <= $key; $i++) {
		$docinfo = <DOCS>;
	    }
	    close(DOCS);
	    ($doctitle, $filename, $offset, $size) = split(/\t/, $docinfo);
	    $selections[$selection_index] = $docinfo;
	    $selection_index++;
	    print "$selection_index.: $doctitle\n";
	    
	}
    }
    
    print "\nView document \# (0 to query again):";
    $selection = <STDIN>;
    if ($selection != 0) {
	($doctitle, $filename, $offset, $size) = split(/\t/, $selections[$selection-1]);
	open(TXT, $filename);
	seek(TXT, $offset, 0);
	read(TXT, $doctxt, $size);
	print $doctxt;
    }

    print "\n\nEnter query: ";
}