Function StringPieceToRune

extern/re2/re2/parse.cc:1391–1413

Removes the next Rune from the StringPiece and stores it in *r. Returns number of bytes removed from sp. Behaves as though there is a terminating NUL at the end of sp. Argument order is backwards from usual Google style but consistent with chartorune.

// Argument order is backwards from usual Google style
// but consistent with chartorune.
static int StringPieceToRune(Rune *r, StringPiece *sp, RegexpStatus* status) {
  // fullrune() takes int, not size_t. However, it just looks
  // at the leading byte and treats any length >= 4 the same.
  if (fullrune(sp->data(), static_cast<int>(std::min(size_t{4}, sp->size())))) {
    int n = chartorune(r, sp->data());
    // Some copies of chartorune have a bug that accepts
    // encodings of values in (10FFFF, 1FFFFF] as valid.
    // Those values break the character class algorithm,
    // which assumes Runemax is the largest rune.
    if (*r > Runemax) {
      n = 1;
      *r = Runeerror;
    }
    if (!(n == 1 && *r == Runeerror)) {  // no decoding error
      sp->remove_prefix(n);
      return n;
    }
  }

  status->set_code(kRegexpBadUTF8);
  status->set_error_arg(StringPiece());
  return -1;
}

// Return whether name is valid UTF-8.
// If not, set status to kRegexpBadUTF8.