#! /n/gnu/bin/gawk -f
# Last edited on 2000-01-31 19:26:50 by stolfi

# Reads a stream of element-factored words, extracts crust+mantle
# prefix and mantle+crust suffix 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 words without core

  gsub(/^[^ktpf]*$/, "", $0);

  if (length($0) == 0) { next; }
  
  # Must have at least one core element:

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

  # Insert "-"s between mantle and core:

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

  # Delete core elements, replace by single "-":

  gsub(/[-].*[-]/, "-", $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;
}