# Script for converting a ToL XML file into a Newick file.
# Use Steve Coile's xmlparser first: 
#    ftp://ftp.freefriends.org/arnold/Awkstuff/xmlparser.awk

# Cam Webb <cam@phylodiversity.net>


BEGIN{
  ORS = "";
  last = "begin NODE";
  depth = -1;
}

{
  gsub(/^\ */,"",$0);

  if (($0 == "begin NODE") && (last == "begin NODE"))
    {
      # down a set of branches
      if (depth >= 0) {print "("};
      depth++;
    }
  else if (($0 == "begin NODE") && (last == "end NODE"))
    {
      # to the next in a polytomy
      print ",";
      last = "begin NODE";
    }
  else if (($0 == "end NODE") && (last == "begin NODE"))
    {
      print label[depth];
      last = "end NODE";
    }
  else if (($0 == "end NODE") && (last == "end NODE"))
    {
      atnode = up[atnode];
      depth--;
      print ")" label[depth];
    }
  else if ($1 == "cdata")
    {
      label[depth] = gensub(/[\(\)\ ,.&;\/]+/, "_", "G", substr($0, 7));
    }

}

END{
  print ";\n";
}