#!/usr/bin/perl

use constant DB     => './irclog.db';
use constant RADIUS => 40;

use strict;
use Lingua::Concordance;
use Text::BarGraph;

# sanity check
my $query = $ARGV[ 0 ];
if ( ! $query ) {

	print "Usage: $0 <regular expression>\n";
	exit;
	
}

# initialize
my $corpus = '';

# create corpus
open INPUT, ' < ' . DB or die "Can't open " . DB . ": $!\n";
while ( <INPUT> ) {

	# clean, parse, and build corpus
	chop;
	my ( $datestamp, $name, $text ) = split /\t/, $_;
	$corpus .= $text . ' ';
	
}
close INPUT;

# initialize concordance
my $concordance = Lingua::Concordance->new;
$concordance->text( $corpus );
$concordance->query( $query );
$concordance->radius( RADIUS );

# do the work
print "Snippets from " . DB . " containing $query:\n";
foreach ( $concordance->lines ) { print "  * $_\n" }
print "\n";

# graph where the query is located in the text
print "A graph illustrating in what percentage of " . DB . " $query is located:\n";
my $barchart = Text::BarGraph->new();
$barchart->autosize( 0 );
$barchart->columns( 40 );
$barchart->sorttype( "numeric" );
$barchart->enable_color( 1 );
print $barchart->graph( $concordance->map );
print "\n";

# done
exit;