Function AddFoldedRange

extern/re2/re2/parse.cc:345–394

Add lo-hi to the class, along with their fold-equivalent characters. If lo-hi is already in the class, assume that the fold-equivalent chars are there too, so there's no work to do.

// If lo-hi is already in the class, assume that the fold-equivalent
// chars are there too, so there's no work to do.
static void AddFoldedRange(CharClassBuilder* cc, Rune lo, Rune hi, int depth) {
  // AddFoldedRange calls itself recursively for each rune in the fold cycle.
  // Most folding cycles are small: there aren't any bigger than four in the
  // current Unicode tables.  make_unicode_casefold.py checks that
  // the cycles are not too long, and we double-check here using depth.
  if (depth > 10) {
    LOG(DFATAL) << "AddFoldedRange recurses too much.";
    return;
  }

  if (!cc->AddRange(lo, hi))  // lo-hi was already there? we're done
    return;

  while (lo <= hi) {
    const CaseFold* f = LookupCaseFold(unicode_casefold, num_unicode_casefold, lo);
    if (f == NULL)  // lo has no fold, nor does anything above lo
      break;
    if (lo < f->lo) {  // lo has no fold; next rune with a fold is f->lo
      lo = f->lo;
      continue;
    }

    // Add in the result of folding the range lo - f->hi
    // and that range's fold, recursively.
    Rune lo1 = lo;
    Rune hi1 = std::min<Rune>(hi, f->hi);
    switch (f->delta) {
      default:
        lo1 += f->delta;
        hi1 += f->delta;
        break;
      case EvenOdd:
        if (lo1%2 == 1)
          lo1--;
        if (hi1%2 == 0)
          hi1++;
        break;
      case OddEven:
        if (lo1%2 == 0)
          lo1--;
        if (hi1%2 == 1)
          hi1++;
        break;
    }
    AddFoldedRange(cc, lo1, hi1, depth+1);

    // Pick up where this fold left off.
    lo = f->hi + 1;
  }
}

// Pushes the literal rune r onto the stack.
bool Regexp::ParseState::PushLiteral(Rune r) {