generate_conf.pl

#!/usr/local/bin/perl

use strict;
use warnings;
use 5.008;

main();

sub main {
    if (@ARGV < 1) {
        print STDERR "Usage: $0 out_dir\n";
        exit 1;
    }
    my $out_dir = shift @ARGV;

    my @C_array = map {2 ** $_} (-4..4);
    my @phi_array = qw(0.000000 0.125661 0.253347 0.385320 0.524401 0.674490 0.841621 1.036433 1.281552 1.644854);

    my $param_and_method = "";

    system("mkdir -p $out_dir") unless -d $out_dir;

    # perceptron
    $param_and_method = '  "method" : "perceptron"';
    generate_conf("$out_dir/perceptron.json", $param_and_method);

    # PA
    $param_and_method = '  "method" : "PA"';
    generate_conf("$out_dir/pa.json", $param_and_method);

    my $i;

    # PA1
    $i = 0;
    foreach my $C (@C_array) {
        $param_and_method  = '  "parameter" : {' . "\n";
        $param_and_method .= '    "regularization_weight" : ' . $C . "\n";
        $param_and_method .= '  },' . "\n";
        $param_and_method .= '  "method" : "PA1"';
        my $n = sprintf("%02d", $i);
        generate_conf("$out_dir/pa1_$n.json", $param_and_method);
        $i += 1;
    }

    # PA2
    $i = 0;
    foreach my $C (@C_array) {
        $param_and_method  = '  "parameter" : {' . "\n";
        $param_and_method .= '    "regularization_weight" : ' . $C . "\n";
        $param_and_method .= '  },' . "\n";
        $param_and_method .= '  "method" : "PA2"';
        my $n = sprintf("%02d", $i);
        generate_conf("$out_dir/pa2_$n.json", $param_and_method);
        $i += 1;
    }

    # CW
    $i = 0;
    foreach my $phi (@phi_array) {
        $param_and_method  = '  "parameter" : {' . "\n";
        $param_and_method .= '    "regularization_weight" : ' . $phi . "\n";
        $param_and_method .= '  },' . "\n";
        $param_and_method .= '  "method" : "CW"';
        my $n = sprintf("%02d", $i);
        generate_conf("$out_dir/cw_$n.json", $param_and_method);
        $i += 1;
    }

    # AROW
    $i = 0;
    foreach my $C (@C_array) {
        $param_and_method  = '  "parameter" : {' . "\n";
        $param_and_method .= '    "regularization_weight" : ' . $C . "\n";
        $param_and_method .= '  },' . "\n";
        $param_and_method .= '  "method" : "AROW"';
        my $n = sprintf("%02d", $i);
        generate_conf("$out_dir/arow_$n.json", $param_and_method);
        $i += 1;
    }

    # NHERD
    $i = 0;
    foreach my $C (@C_array) {
        $param_and_method  = '  "parameter" : {' . "\n";
        $param_and_method .= '    "regularization_weight" : ' . $C . "\n";
        $param_and_method .= '  },' . "\n";
        $param_and_method .= '  "method" : "NHERD"';
        my $n = sprintf("%02d", $i);
        generate_conf("$out_dir/nherd_$n.json", $param_and_method);
        $i += 1;
    }

    # SCW1
    $i = 0;
    foreach my $phi (@phi_array) {
        foreach my $C (@C_array) {
            $param_and_method  = '  "parameter" : {' . "\n";
            $param_and_method .= '    "regularization_weight" : ' . $phi . ",\n";
            $param_and_method .= '    "penalty_weight" : ' . $C . "\n";
            $param_and_method .= '  },' . "\n";
            $param_and_method .= '  "method" : "SCW1"';
            my $n = sprintf("%02d", $i);
            generate_conf("$out_dir/scw1_$n.json", $param_and_method);
            $i += 1;
        }
    }

    # SCW2
    $i = 0;
    foreach my $phi (@phi_array) {
        foreach my $C (@C_array) {
            $param_and_method  = '  "parameter" : {' . "\n";
            $param_and_method .= '    "regularization_weight" : ' . $phi . ",\n";
            $param_and_method .= '    "penalty_weight" : ' . $C . "\n";
            $param_and_method .= '  },' . "\n";
            $param_and_method .= '  "method" : "SCW2"';
            my $n = sprintf("%02d", $i);
            generate_conf("$out_dir/scw2_$n.json", $param_and_method);
            $i += 1;
        }
    }
}

sub generate_conf {
    my ($out_f, $param_and_method) = @_;

    my $doc = <<"EOD";
{
  "converter" : {
    "string_filter_types" : {},
    "string_filter_rules" : [],
    "num_filter_types" : {},
    "num_filter_rules" : [],
    "string_types" : {},
    "string_rules" : [
      { "key" : "*", "type" : "str", "sample_weight" : "bin", "global_weight" : "bin" }
    ],
    "num_types" : {},
    "num_rules" : [
      { "key" : "*", "type" : "num" }
    ]
  },
$param_and_method
}
EOD

   open(my $ofh, '>', $out_f) or die $!;
   print {$ofh} $doc;
   close($ofh);

}