# call program as so:
# cat tagger_output.tts | perl convert.pl > YOURFILE_pos_level.xml

# Before calling the program, it might be a good idea to first do the
# following:
# mv YOURFILE_pos_level.xml backup.xml

# Do not add extra newlines - apparently, this messes MMAX2 up (?)
$header = '<?xml version="1.0" encoding="US-ASCII"?>'."\n".'<!DOCTYPE markables SYSTEM "markables.dtd">'."\n".'<markables xmlns="www.eml.org/NameSpaces/POS">'."\n";

sub make_line {
  $sid=$_[0]; $pos_tag=$_[1];
  $line = '';
  $line = '<markable ';
  $line .= 'mmax_level="pos" ';
  $line .= 'id="markable_'.$sid.'" ';
  $line .= 'span="word_'.$sid.'" ';
  $line .= 'tag="'.$pos_tag.'" ';
  $line .= '/>';
  $line .= "\n";
  return $line
}

print $header;

$index = 1;
while (<>) {
  chomp;
  # ignore lines starting with '%%', i.e., comment lines from TnT
  if (!/^%%/) {

    # split the line, in order to get the POS tag
    @my_list = split(/\s+/,$_);
    # ignore the word & grab just the POS tag (fyi: this assumes that
    # each word has no spaces in it)
    $pos_tag = $my_list[1];

    # make the appropriate XML line & print it out
    print &make_line($index,$pos_tag);
    # iterate the index properly
    $index++;
  }
}

# If MMAX2 gives weird behavior, it may help to try with/without a
# trailing newline here (?)
print "</markables>";