parsehtml

alias dehtml {
  var %commonchars = lt|gt|amp|quot|apos
  var %commoncharcodes = 60|62|38|34|39
  var %i = $regex($1, /(&(?:#\d+| $+ %commonchars $+ );)/gi), %chr, %after, %retval
  if (!%i) { return $1 }
  set -n %retval $1
  while (%i) {
    %chr = $mid($regml(%i), 2, -1)
    ; some very common escapes
    if ($findtok(%commonchars, %chr, 1, 124)) { %chr = $gettok(%commoncharcodes, $ifmatch, 124) }
    ; since mIRC doesn't support unicode characters, turn them into question marks
    else {
      %chr = $mid(%chr, 2)
      if (%chr > 255) { %chr = 63 }
    }

    set -n %after $mid(%retval, $calc($regml(%i).pos + $len($regml(%i))))
    set -n %retval $+($left(%retval, $calc($regml(%i).pos - 1)), $chr(%chr), %after)
    dec %i
  }
  return %retval
}
alias htmlfree {
  var %x, %i = $regsub($1-,/(^[^<]*>|<[^>]*>|<[^>]*$)/g,$null,%x), %x = $remove(%x,&nbsp;)
  return $dehtml(%x)
}