#! /usr/bin/gawk -f
# Last edited on 2026-07-23 18:13:28 by stolfi
# Sampling functions for engp/mky
# Pseudo-English generated by a Markov chain
# Words should be already all lowercase, with no hyphens or apostropes,
# so we don't need to do anything.
# To be included in wds_to_tlw.gawk

function smp_define_patterns(smp,sec)
{
  # No patterns needed 
  if (sec != "tot.1") { data_error(("invalid section \"" sec "\"")); }
}

function smp_reclassify_token(smp,sec,cursec,curlin,type,wd)
{
  # Reject subsections that are not main text:
  if (cursec !~ /{v[0-9]+}{tx} *$/) { return "x"; }
  
  # Discard punctuation other than parag breaks: 
  if ((type == "p") && ( wd != "=")) { return "n"; }
  
  return type;
}

function smp_fix_token(smp,sec,type,wd)
{
  # Nothing to do:
  return wd;
}

function smp_is_good_token(smp,sec,type,wd)
{ 
  # Accept only lowercase alpha (no apostrophes or hyphens).
  return (wd ~ /^[a-z]+$/);
}