function decode_html_entity()

in src/_Private/decode_html_entity.php [17:55]

• 35 lines of code
• 9 McCabe index (conditional complexity)

function decode_html_entity(string $string): ?(string, string, string) {
  if ($string[0] !== '&') {
    return null;
  }

  $matches = Regex\first_match(
    $string,
    re"/^&(#[0-9]{1,8}|#X[0-9a-f]{1,8}|[a-z][a-z0-9]*);/i",
  );
  if ($matches is null) {
    return null;
  }
  $match = $matches[0];

  $table = get_html_entity_table();
  $out = $table[$match] ?? null;
  if ($out !== null) {
    return tuple($match, $out, Str\strip_prefix($string, $match));
  }

  if ($match[1] !== '#') {
    return null;
  }

  $out = \html_entity_decode(
    $match,
    /* HH_FIXME[4106] */ /* HH_FIXME[2049] */ \ENT_HTML5,
    'UTF-8',
  );
  if ($out === $match) {
    return null;
  }
  if ($out === "\000") {
    $out = UNICODE_REPLACEMENT_CHARACTER;
  } else if (!\mb_check_encoding($out, 'UTF-8')) {
    $out = UNICODE_REPLACEMENT_CHARACTER;
  }
  return tuple($match, $out, Str\strip_prefix($string, $match));
}