#!/usr/bin/perl

$usage = "usage: cat <parcFiles> | $0 <wsjDir>

where <wsjDir> contains the files wsj_2300.mrg, wsj_2301.mrg, ..., wsj_2399.mrg
      from section 23 of the UPenn WSJ Treebank-II (LDC catalog no.: LDC95T7),
      <parcFiles> are the PARC 700 dependency bank files.

";

($wsjDir) =  shift @ARGV || die $usage;

while(<STDIN>){	
    chomp;
    if(($file_id,$sent_id,$parc_id) = $_ =~ /\_(\d+)\.(\d+).*\.(\d+)/){ 	
	open(TREES, "<$wsjDir/wsj_$file_id.mrg");
	read TREES,$data,1000000;
	@D = split(/(?=\( \([A-Z])/, $data);
    	close(TREES);
	print "parc \#$parc_id\, wsj file wsj_$file_id.mrg, sentence \#$sent_id: \n\n";
	print $D[$sent_id], "\n\n";
    }
}