#!/usr/bin/perl

use strict;
use XML::TMX::Reader;
use libmorfo_perl;

use Analysis;

my $FREELINGDIR = '/home/jorge/tools/';
my $DATA = $FREELINGDIR . 'share/FreeLing/';


my $l1="pt";
my $l2="en";


my %analysers = ();

#sub apt{ uc($_[0])}
#sub aen{ uc($_[0])}

for my $file (@ARGV){
  my $tm = XML::TMX::Reader->new($file);

	my @langs = $tm->languages;

	# initialize analysers
	map { $analysers{$_} = init($_) if not $analysers{$_}; } @langs;



  #if (not defined $l1) {
  #  ($l1,$l2) = sort ($tm->languages);
  #  print "Using languages: $l1/$l2\n"
  #}

  $tm->for_tu(
      { output => "$file._" },
      sub {
         my $tu = shift;

				 $tu->{$l1} = analyse($l1, $tu->{$l1})->to_xml;
				 $tu->{$l2} = analyse($l2, $tu->{$l2})->to_xml;

         $tu;
      });
}



sub init {
	my $LANG = shift;
	my $analyser = {};

	my $tokenizer = libmorfo_perl::tokenizer->new("$DATA$LANG/tokenizer.dat");
	my $splitter = libmorfo_perl::splitter->new("$DATA$LANG/splitter.dat");

	my $options = libmorfo_perl::maco_options->new($LANG);
	$options->set_active_modules(1,1,1,1,1,1,1,1,0,0);
	$options->set_data_files(
	    "$DATA$LANG/locucions.dat",
			"$DATA$LANG/quantities.dat",
			"$DATA$LANG/afixos.dat",
			"$DATA$LANG/probabilitats.dat",
			"$DATA$LANG/maco.db",
			"$DATA$LANG/np.dat",
			"$DATA/common/punct.dat",
			"$DATA$LANG/corrector/corrector.dat");

	my $maco = libmorfo_perl::maco->new($options);
	my $tagger = libmorfo_perl::hmm_tagger->new($LANG,"$DATA$LANG/tagger.dat",1,2);

	$analyser->{tokenizer} = $tokenizer;
	$analyser->{splitter}  = $splitter;
	$analyser->{maco}      = $maco;
	$analyser->{tagger}    = $tagger;

	return $analyser;
}

sub analyse {
	my $LANG = shift;
	my $str  = shift;

	my $words     = $analysers{$LANG}{tokenizer}->tokenize($str);
	my $sentences = $analysers{$LANG}{splitter}->split($words,1);

	# análise morfológica
	$sentences = $analysers{$LANG}{maco}->analyze($sentences);
	# PoS tagging
	$sentences = $analysers{$LANG}{tagger}->analyze($sentences);

	my $analysis = Analysis->new();
	$analysis->{sentences} = $sentences;
	
	return $analysis;
}