#!/usr/bin/perl -s

my $id= shift or die;

our($oco,$cor,$dup,$equiv,$debug,$requiv);
use NAT::Client;
use Data::Dumper;
use strict;

#my $client = NAT::Client->new( LocalDumper => "$id" );
my $client = NAT::Client->new( Local => "$id");

my $minimum_oco = $oco|| 20;
my $minimum_cor = $cor|| 0.70;
my %mem;

$client->iterate( 
  {Language =>'source'},
  sub { my %p = @_;
        my $bestkey = (sort {$p{trans}{$b} <=> $p{trans}{$a}} keys %{$p{trans}})[0];
        return if $p{trans}{$bestkey} < $minimum_cor;
        return if $p{count}           < $minimum_oco;
        return if $p{word} eq $bestkey;
        return if $p{word} =~ /unicode/;
        return if $bestkey =~ /unicode/;
        return if $p{word} =~ /\d/;
        return if $bestkey =~ /\d/;
        if($dup)  {print "$p{word} $bestkey\n"}
        if($debug){print "$p{word} $bestkey=$p{count}=$p{trans}{$bestkey}\n"}
        $mem{$p{word}}=$bestkey;
      });

print "==============================================\n";
$client->iterate( 
  {Language =>'target'},
  sub { my %p = @_;
        my $bestkey = (sort {$p{trans}{$b} <=> $p{trans}{$a}} keys %{$p{trans}})[0];
        return if $p{trans}{$bestkey} < $minimum_cor;
        return if $p{count}           < $minimum_oco;
        return if $p{word} eq $bestkey;
        return if $p{word} =~ /unicode/;
        return if $bestkey =~ /unicode/;
        return if $p{word} =~ /\d/;
        return if $bestkey =~ /\d/;
        if($dup)  {print "$p{word} $bestkey\n"}
        if($debug){print "$p{word} $bestkey=$p{count}=$p{trans}{$bestkey}\n"}
        if($equiv  and $p{word} eq $mem{$bestkey}){print "$bestkey $p{word}\n"};
        if($requiv and $p{word} eq $mem{$bestkey}){print "$p{word} $bestkey\n"};
      });



__END__

#!/usr/bin/perl -w

use strict;

use NAT;
use NAT::Dict;
use Data::Dumper;

my $dict_name = shift;
unless ($dict_name =~ m!\.db!) {
  die 'de momento so funciono com ficheiros dbs';
}

my $dict = NAT::Dict::open(dbfile => $dict_name);
__END__

=head1 NAME

jj-trans-equiv - extract word equivalents from PTD

=head1 SYNOPSIS

 jj-trans-equiv -equiv   corpus-dir    (equiv source-traget)
 jj-trans-equiv -requiv  corpus-dir    (reverse target-source)
 jj-trans-equiv -dup  corpus-dir       (source-target + traget-source)

=head1 DESCRIPTION

=head1 Options

 -cor=0.6  minimum percentage of the best word translation (0.7)
 -oco=10   minimum number of ocurrences of the source word (20)

=head1 AUTHOR

J.Joao Almeida, jj@di.uminho.pt

=head1 SEE ALSO

perl(1).

=cut