#!/usr/bin/perl

use strict;
use warnings;
use utf8;
use Encode;

use Lingua::FreeLing;
use Lingua::FreeLing::Simple;

my %analysers = ();

my @langs = qw|pt|;

# initialize analysers
map { $analysers{$_} = new Lingua::FreeLing::Simple($_) if not $analysers{$_}; } @langs;

#				Encode->from_to($tu->{$l1}, "utf8", "iso-8859-1");
#				Encode->from_to($tu->{$l2}, "utf8", "iso-8859-1");
				
#				$tu->{$l1} = to_string($analysers{$l1}->analyse($tu->{$l1}));
#				$tu->{$l2} = to_string($analysers{$l2}->analyse($tu->{$l2}));

#				Encode->from_to($tu->{$l1}, "iso-8859-1", "utf8");
#				Encode->from_to($tu->{$l2}, "iso-8859-1", "utf8");

binmode(STDOUT, ":utf8");

my $str1 = "mudanças é que é bonito.";
my $str2 = $str1;

##Encode::decode_utf8(Foo::Bar::escape_html(Encode::encode_utf8($what)));


##Encode::from_to($str2, "utf8", "iso-8859-1");

#Encode->from_to($str2, "utf8", "iso-8859-1");

my $s1 = $analysers{"pt"}->analyse($str1);

 $str2 = encode("iso-8859-1", $str2);

print "'$str1'\n";
#print "'$str2'\n";
my $s2 = $analysers{"pt"}->analyse($str2);


print to_string($s1);
print "----------\n";
print to_string($s2);

sub to_string {
	my $sentences = shift;
	my $xml = "";
	my $template_word = qq|\t#FORM# #LEMMA# #TAG#|;

	for my $s (@$sentences) {
		my $sentence = "";

		my $ws = $s->get_words;
		for my $w (@$ws) {
			my $form  = decode("iso-8859-1", $w->get_form);
			my $lemma = decode("iso-8859-1",$w->get_lemma);
			my $tag   = decode("iso-8859-1",$w->get_parole);


			my $word = $template_word;
			$word =~ s/#FORM#/$form/ge;
			$word =~ s/#LEMMA#/$lemma/ge;
			$word =~ s/#TAG#/$tag/ge;

			$sentence.= "$word\n";
		}

		$sentence.= "";
		$xml.= $sentence;	
	}

	return $xml;
}

sub to_xml {
	my $sentences = shift;
	my $xml = "";
	my $template_word = qq|<word form="#FORM#" lemma="#LEMMA#" tag="#TAG#" />|;

	for my $s (@$sentences) {
		my $sentence = "\t<sentence>\n";

		my $ws = $s->get_words;
		for my $w (@$ws) {
			my $form  = $w->get_form;
			my $lemma = $w->get_lemma;
			my $tag   = $w->get_parole;


			my $word = $template_word;
			$word =~ s/#FORM#/$form/ge;
			$word =~ s/#LEMMA#/$lemma/ge;
			$word =~ s/#TAG#/$tag/ge;

			$sentence.= "\t\t$word\n";
		}

		$sentence.= "\t</sentence>\n";
		$xml.= $sentence;	
	}

	return $xml;
}