2num.pl

#!/opt/local/bin/perl
##
## Slightly more idiomatic and optimized Perl of 2num.pl from
## http://github.com/brendano/awkspeed/tree/master
##
use strict;
use warnings;

open( my $vocab_fh, '>', "vocab" ) or die "vocab: $!\n";

my %jmap;
my %i_file;
my $j_ctr = 0;

for my $infile ( @ARGV ) {
  print STDERR "new file $infile\n";
  open( my $in_fh, '<', $infile ) or die "$infile: $!\n";
  open( my $out_fh, '>', "${infile}n" ) or die "${infile}n: $!\n";

  my %imap;
  my( $ival, $jval );
  while( <$in_fh> ) {
    my( $item, $feat, $val ) = split( / / );
    $ival = ($imap{$item} ||= ++$i_file{ $infile });

    print $vocab_fh qq{$feat\n} unless exists $jmap{ $feat };
    $jval = ($jmap{$feat} ||= ++$j_ctr);

    print $out_fh qq{$ival $jval $val};
  }
}

__END__