#!/usr/bin/perl  -s

our ($id);

$id||="__jj1-constituicao";

my @l = ();
my @inv =(direction => "<~");

use NAT::Client;
use Data::Dumper;
$Data::Dumper::Indent=1;
$Data::Dumper::Terse=1;

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

while(<>){
  chomp;
  my %r = ();
  my $ws = split(/\s+/,$_);
  my $word = $_;

  print "\n";
## pre ngrams
  if ($ws==3){
    for (sngrams({max=>10},"* $word")){
      next unless $_;
      printf("%-13s |",  "$_->[4] $_->[0]");
      print "\n";
    }
  } 
  elsif($ws == 2){
    for (sngrams({max=>10},"* $word")){
      next unless $_;
      $tmp = "";
      for (sngrams({max=>10},"* $_->[0] $word")){
        next unless $_;
        $tmp .= "$_->[0] |";
      }
      printf("%85s| %-10s\n", substr($tmp,0,85), "$_->[3] $_->[0]");
    }
  }
  else {
    for (sngrams({max=>10},"* $word")){
        next unless $_;
        $tmp = "";
        for (sngrams({max=>10},"* * $_->[0] $word")){
          next unless $_;
          $tmp .= "$_->[0] $_->[1] |";
        }
        printf("%85s| %-10s\n", substr($tmp,0,85), "$_->[2] $_->[0]");
    }
  }

  print "===== $word =====\n";

  if ($ws==3){
    for (sngrams({max=>10},"$word *")){
      next unless $_;
      printf("%-13s |",  "$_->[4] $_->[3]");
      print "\n";
    }
  } 
  elsif($ws == 2){
    for (sngrams({max=>10},"$word *")){
      next unless $_;
      printf("%-13s |",  "$_->[3] $_->[2]");
      for (sngrams({max=>10},"$word $_->[2] *")){
        next unless $_;
        print ("| $_->[3] ");
      }
      print "\n";
    }
  }
  else {
    for (sngrams({max=>10},"$word *")){
        next unless $_;
        printf("%-13s |",  "$_->[2] $_->[1]");
        for (sngrams({max=>10},"$word $_->[1] * *")){
          next unless $_;
          print ("| $_->[2] $_->[3]");
        }
        print "\n";
    }
  }
  print "\n";
}

sub sngrams{
 my %opt =(max => 50);
 if(ref($_[0]) eq "HASH") {%opt = (%opt , %{shift(@_)}) } ;
 my $exp=shift;
 my $ng=split(/\s+/,$exp);

 if($opt{max}){  # grep {$_} 
    ((sort {$b->[$ng]<=>$a->[$ng]} @{$client->ngrams($exp)})[0..$opt{max}-1])
 }
 else { (sort {$b->[$ng]<=>$a->[$ng]} @{$client->ngrams($exp)}) }
}

__END__

=encoding utf8

=head1 NAME

jj-9 - compact tetragrams from 1 or 2 words Pro e pre

=head1 SYNOPSIS

=head1 DESCRIPTION

(Previously: jj-1 -> to create the corpus)

Print compact tetragrams . Example of output:


 itleart audiências |titleart decisões |ii org| 24 dos    
 |mandato perante |governo e |judiciais são |3| 16 os     
 justiça incumbe |- compete |disposto quanto || 5 aos     
 egorias |a existência |seguintes categorias || 3 de      
                                  juízes dos || 3 restantes
                                              | 2 titleart
                                              | 1 constituídos
                                     - podem || 1 existir 
                             conjuntamente , || 1 em      
                         tribunais marítimos || 1 ,       
 ===== tribunais =====
 10 judiciais  || de primeira| de segunda| , sem| e o| são o
 6 de          || conflitos .| qualquer instância| primeira i
 6 administrativos || e fiscais
 4 são         |
 3 da          || relação .| relação e
 3 .           |
 3 :           |
 3 militares   || , é| com competência
 2 e           || do ministério| os demais
 2 que         || não sejam| apliquem norma


In order to be accurate a big corpus is necessary...

=head1 AUTHOR

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

=head1 SEE ALSO

perl(1).

=cut