#! /usr/bin/gawk -f
# Last edited on 2025-09-24 15:05:49 by stolfi
# Sampling functions for latn/ahl
# Transcription of the so-called "Alchemical herbal" by Marco Ponzi.
# To be included in wds-to-tlw

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

function smp_is_good_token(smp,sec,type,wd)
{ 
  # Accept only lowercase alpha.
  # The text uses neither hyphen nor apostrophe. 
  return (wd ~ /^[a-z]+$/);
}

function smp_reclassify_token(smp,sec,cursec,curlin,type,wd)
{
  # Delete anything outside section {sec}:
  if (sec == "tot.1")
    { if (cursec !~ /^{B}{h[0-9]+}/) { return "n"; } }
  else  
    { arg_error(("bad subsection \"" subsec "\"")); }
  
  # Delete titles:
  if (cursec ~ /{tt} *$/) { return "n"; }
  
  # Reject any subsection that is not prose text:
  if (cursec !~ /{p[0-9]+} *$/) { return "x"; };
  
  # Discard punctuation other than parag breaks: 
  if ((type == "p") && (wd != "=")) { return "n"; }

  return type;
}

function smp_fix_token(smp,sec,type,wd)
{
  # Map upper case to lower case (just in case).
  wd = tolower(wd);
  return wd;
}