# Last edited on 2005-01-08 15:02:44 by stolfi

The input to the SELVA grammar reducer and reorderer
is a list of rules, one per line, of the form "{H} -> {L} {R}"
where {H}, {L}, {R} are strings matching the pattern
[*$_(){},A_Za-z0-9]+.  The part {R} or both {L} and {R}
may be omitted. 

The output is a reduced grammar in numeric format,
suitable for the SELVA parser (cfp). 

  begin grammar (format of 2004-11-01)
  symbols = {NS}
  rules = {NR}
  {IS}: {SNAME}
  ...
  
  {IR}: {IH} {IL} {IR} # {H} -> {L} {R}
  ...
  end grammar
 
where

  {NS} and {NR} are the symbol and rule counts;

  {IS} and {IR} are sequential symbol and rule indices,
  in sequence (ranging from 1 to {NS} or {NR});
  
  {SNAME} is a symbol's name (a string of nonblank chars);
  
  {IH}, {IL}, {IR} are the left-hand symbol and the 
  two parts of the right-hand side of a rule.
  The symbol {IL} is 0 for unary rules.
  
  {H} -> {L} {R} is the rule in symbolic form.