#!/usr/bin/perl -s

use warnings;
use strict;
use XML::DT;

our ($h);

print usage() if $h;

my $filename;
while ($filename = shift) {

    my %file;
    my %seg;
    my $i = 0;

    print STDERR "Processing $filename..";
    my %handler = (
                   -default => sub{ "" },

                   seg => sub{ "$c" },
                   tuv => sub{
                       my $fd;
                       $v{lang} = $v{'xml:lang'} if exists($v{'xml:lang'});
                       print STDERR "." unless ($i++%500);
                       $seg{$v{lang}}++;
                       unless (exists($file{"$filename-$v{lang}"})) {
                           my $x;
                           open $x, ">:utf8", "$filename-$v{lang}" or die "Cannot open $filename-$v{lang}";
                           $file{"$filename-$v{lang}"} = $x;
                       }
                       $fd = $file{"$filename-$v{lang}"};
                       myprint($fd, $c);
                   },
                  );

    dt($filename, %handler);
    print STDERR " done\n";

    for (keys %seg) {
        print STDERR "$_: $seg{$_} segments\n";
    }
    for (keys %file) {
        close $file{$_};
    }
}

sub myprint{
    my ($f,$tu) = @_;
    for ($tu){
        s/<.*?>/ /gs;
        s/[\|\$]/ /gs;
        s/(\w)([.;,!:?«»"])/$1 $2/g;
        s/([.;,!:?«»"])(\w)/$1 $2/g;
        s/\s\s+|^\s+|\s+$/ /g;
    }
    print {$f} "$tu\n\$\n";
}


sub usage {
    print "nat-tmx2pair: splits a TMX file into several files, one for each language\n\n";
    print "\tnat-tmx2pair <file.tmx>\n\n";
    print "For more help, please run 'perldoc nat-tmx2pair'\n";
    exit (0);
}

__END__

=encoding UTF-8

=head1 NAME

nat-tmx2pair - splits a TMX file into several files, one for each language

=head1 SYNOPSIS

 nat-tmx2pair f.tmx

=head1 DESCRIPTION

This script takes a TMX file and outputs n different files, one for
each language with translation units separated by dollar signs
(NATools standard input format).

The files creates are based on the tmx filename, with the language tag
appended in the end.

=head1 SEE ALSO

NATools documentation, perl(1).

=head1 AUTHOR

Alberto Simões, E<lt>ambs@cpan.orgE<gt>

=head1 COPYRIGHT AND LICENSE

Copyright (C) 2006-2012 by Alberto Manuel Brandão Simões

=cut