# Last edited on 2025-09-24 19:31:01 by stolfi
# Sampling functions for engp/cul
# English - Plant names from Culpeper's herbal, mapped to lowercase.
# To be included in select-evt-lines, fix_sample_tokens.gawk, select-gud-bad-words

function select_evt_line(subsec,chapter,unit,linenum)
{
  # Take part d (list of plants by planets), chapter 1, unit type "V" (plantlist);
  # also part h (herbal proper), all chapters, unit type "T" (chapter title).
  # Make it all into a single subsection.
  if (subsec = "bod.1")
    { 
      return \
        ((chapter ~ /^[d]/) && (unit ~ /^[V]/)) || \
        ((chapter ~ /^[h]/) && (unit ~ /^[T]/));
    }
  else
    { arg_error(("bad subsection \"" subsec "\"")); }
}

function fix_raw_word(word)
{
  # Map upper case to lower case, break at hyphens.
  word = tolower(word);
  if (word !~ /^[-]*$/) { gsub(/[-]/, "\n", word); }
  return word;
}

function define_patterns()
{
  # No patterns needed 
}

function is_good_word(word)
{ 
  # Accept only lowercase alpha, plus apostrophe.
  # Apostrophes can't be doubled.
  # Note that 'tis OK to begin an' end with apostrophe! 
  return (word ~ /^([']?[a-z])+[']?$/);
  #
  # The following allows hyphenated words.
  # Note that each word of 
  # an hyphenated compound must contain at least one letter.
  # return (word ~ /^(([']?[a-z])+[']?)([-]([']?[a-z])+[']?)*$/);
}