#! /n/gnu/bin/gawk -f
# Last edited on 2000-05-17 09:11:48 by stolfi

# Reads a stream of element-factored words, extracts full core+mantle
# and sufcrust separated by "-".

/^ *$/{next;}
/./{ 
  
  # Delete dummy (empty) factors:

  gsub(/{[_]*}/, "", $0);
  gsub(/[_][_]*/, "", $0);

  # Map "ch" and "sh" to "C" and "S" to simplify processing:
  
  gsub(/ch/, "C", $0);
  gsub(/sh/, "S", $0);

  # Delete leading non-mantle, non-core elements:

  gsub(/^[^CSktpfech]*[{]/, "{", $0);
  gsub(/^[^CSktpfech]*$/, "", $0);
  
  if (length($0) == 0) { next; }
  
  # Must have at least one element:

  if ($0 !~ /^[{].*[CSktpfech].*[}]/)  { error(("bad format = «" $0 "»")); }

  # Insert "-" between core-mantle and suffix:

  $0 = gensub(/[}]([^CSktpfech]*)$/, "}-\\1", "g", $0); 

  # Delete braces
  
  gsub(/[{}]/, "", $0);

  # Restore "cs" and 'sh":
  
  gsub(/C/, "ch", $0);
  gsub(/S/, "sh", $0);

  print $0;
}

function error(msg)
{ 
  printf "line %d: %s\n", NR, msg >> "/dev/stderr";
  abort = 1;
  exit 1;
}