#!/usr/bin/perl

use strict;
use warnings;
use utf8;

use XML::TMX::Reader;
use XML::TMX::Writer;
use Lingua::FreeLing::Simple;

use Data::Dumper;

my %analysers = ();

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

	my @langs = $tmin->languages;

	$tmout->start_tmx(DATATYPE => 'xml', OUTPUT => "$file.freeling");

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

	# process TMX file
  $tmin->for_tu(
      { },
      sub { my $tu = shift;
				my %tus = ();
				for my $lang (@langs) {
					if (defined $tu->{$lang}) {
						my $sentences = $analysers{$lang}->analyse($tu->{$lang});
						$tus{$lang} =
							join "\n", Lingua::FreeLing::Simple::forall($sentences, { sentence => \&to_xml , returns => 1});
					}
				}
				$tmout->add_tu(%tus);
      });
	
	$tmout->end_tmx();
}


sub to_xml {
	my $sentence = shift;

	my $ws = $sentence->get_words; 

	my @words = Lingua::FreeLing::Simple::get_tuple($ws, { lemma => 1, tag => 1 });

	"<sentence>\n\t" .
		(join "\n", (map { my $w = $_; $_ = ("<word " . (join " ", (map { $_ = qq|$_="$w->{$_}"| } (keys %$w))) . " />") } @words))
		. "<sentence>\n";
}