#! /usr/bin/gawk -f
# Last edited on 2025-05-09 23:17:23 by stolfi

# Reads a file where each non-blank line has the format "{LOC} {XEVALIN}" where {LOC}
# is an EVT locator, like "<f85v2.B2.8;V>", and {XEVALIN} is a Voynichese text line in the 
# XEVA encoding.
#
# Scans {XEVALIN} looking for XEVA elements that are "(*)" or 
# are not single basic EVA codes ("%", "a", ... "y", "Ch", "Sh", "CTh", etc) in parens. In particular, 
# any elems that contain weirdo codes "&NNN" or more than one basic EVA glyph.
# 
# For each such element {XEL} found, outputs a line "<LOC> {LEFT}¦{XEL}¦{RIGHT}" where {LEFT} and {RIGHT} are the truncated left and right contexts in {XEVALIN}.

BEGIN{ abort = -1; }

(abort >= 0) { exit(abort); }

/^</ {
  if (NF != 2) { prog_error("NF"); }
  loc = $1; xlin = $2;
  left = ""; 
  while (xlin != "") {
    if (match(xlin, /^[-=.,]/)) {
      left = (left subtr(xlin, 1, 1));
      xlin = substr(xlin, 2);
    } else if (match(xlin, /^[(][^()]+[)]/)) {
      if (RSTART != 1) { prog_error("RSTART"); }
      xel = sibstr(xlin, 1, RLENGTH);
      xlin = substr(xlin, RLENGTH + 1);
      if (! match(xel,