#!/usr/local/bin/perl

# This script will generate  word level networks for the word-pair test.
# usage: wp_net.prl wp_gram_file nn nl
#
# TL 7/1998
#

$[ = 1;						# set array base to 1

if ( $#ARGV != 4 ) {
	die "usage: wp_net.prl word-pair_grammar_file output_net_file nn nl\n";
}

$wp_file = $ARGV[1];
if ( ! -f $wp_file ) {
	die "word-pair_grammar_file $wp_file not found.\n";
}

$net_file = $ARGV[2];
$nn = $ARGV[3];
$nl = $ARGV[4];

open(NETFILE,">$net_file") or die "Can't open file $net_file\n";

for ( $i=1; $i<=2; $i++ ) {
	open(WPFILE,"<$wp_file") or die "Can't open file $wp_file\n";
	while (<WPFILE>) {
	    chop;				# strip record separator
		/\*/ && next;		# skip comment line
	    @Fld = split(' ', $_, 9999);
	
		if ($#Fld == 0) {
			if ($ln > 990) {
			    $links = 1;
			}
			elsif (!$ln) {
			    printf NETFILE "N=%d L=%d\n", $nn, $nl;
			    $stwd = '!SENT_START';
			    printf NETFILE "I=%-4d W=%s\n", 0, $stwd;
			    $enwd = '!SENT_END';
			    $word = 'SENTENCE-END';
			    $num{$word} = 1;
			    $words{1} = $word;
			    printf NETFILE "I=%-4d W=%s\n", 1, $enwd;
			    $ln = 2;
			}
		}
		elsif ($links) {
			if (substr($Fld[1], 1, 1) eq '>') {
			    $word = substr($Fld[1], 2, length($Fld[1]) - 1);
			}
			else {
			    $like = -log($nfoll{$word});
			    if ($word eq 'SENTENCE-END') {
					printf NETFILE "J=%-5d S=%-4d E=%-4d l=%.4f\n", $la, 0, $num{$Fld[1]}, $like;
			    }
			    else {
					printf NETFILE "J=%-5d S=%-4d E=%-4d l=%.4f\n", $la, $num{$word}, $num{$Fld[1]}, $like;
			    }
			    $la++;
			}
		}
		elsif ($ln) {
			if (substr($Fld[1], 1, 1) eq '>') {
			    $word = substr($Fld[1], 2, length($Fld[1]) - 1);
			    $nfoll{$word} = 0;
			    if ($num{$word}) {
					$n = $num{$word};
			    }
			    else {
					$n = $ln;
					$num{$word} = $ln++;
					printf NETFILE "I=%-4d W=%s\n", $n, $word;
			    }
			}
			else {
			    $nfoll{$word}++;
			}
		}
	}
}

close(WPFILE);
close(NETFILE);