#!/usr/local/bin/perl -w
use Getopt::Std;

my $maxsize = 500_000;
my $maxsizemark = '500K';

sub usage { <<EOF
Usage: $0 stem
	Where 'stem' is like 'ecaux/1M-full-train-25'
EOF
}

my %opt;
getopts('h', \%opt);

@ARGV || die usage();
$opt{h} && die usage();


my $stem = shift;
my $arch = $ENV{ARCH};
my $c45 = "/u/dpb/nlp/c4.5-$arch/Src/c4.5";

my $size = `wc -l $stem.data`;
$size =~ s/\s*(\d+).*/$1/;
my $sizemark = $size > $maxsize? $maxsizemark: '';
my $mstem = $stem;
$mstem =~ s/^(.*\/)(.*)/$1 . $sizemark . $2/e;

system("$c45 -n $maxsize -f $stem > $mstem.output");
system("rm $stem.unpruned");
if ($stem ne $mstem) {
  system("mv $stem.tree $mstem.tree");
  system("ln -s $stem.names $mstem.names");
}
