Method ParseCharClass

extern/re2/re2/parse.cc:1907–2012

Parses a possibly-negated character class expression like [^abx-z[:digit:]]. Sets *s to span the remainder of the string. Sets *out_re to the regexp for the class.

// Sets *s to span the remainder of the string.
// Sets *out_re to the regexp for the class.
bool Regexp::ParseState::ParseCharClass(StringPiece* s,
                                        Regexp** out_re,
                                        RegexpStatus* status) {
  StringPiece whole_class = *s;
  if (s->empty() || (*s)[0] != '[') {
    // Caller checked this.
    status->set_code(kRegexpInternalError);
    status->set_error_arg(StringPiece());
    return false;
  }
  bool negated = false;
  Regexp* re = new Regexp(kRegexpCharClass, flags_ & ~FoldCase);
  re->ccb_ = new CharClassBuilder;
  s->remove_prefix(1);  // '['
  if (!s->empty() && (*s)[0] == '^') {
    s->remove_prefix(1);  // '^'
    negated = true;
    if (!(flags_ & ClassNL) || (flags_ & NeverNL)) {
      // If NL can't match implicitly, then pretend
      // negated classes include a leading \n.
      re->ccb_->AddRange('\n', '\n');
    }
  }
  bool first = true;  // ] is okay as first char in class
  while (!s->empty() && ((*s)[0] != ']' || first)) {
    // - is only okay unescaped as first or last in class.
    // Except that Perl allows - anywhere.
    if ((*s)[0] == '-' && !first && !(flags_&PerlX) &&
        (s->size() == 1 || (*s)[1] != ']')) {
      StringPiece t = *s;
      t.remove_prefix(1);  // '-'
      Rune r;
      int n = StringPieceToRune(&r, &t, status);
      if (n < 0) {
        re->Decref();
        return false;
      }
      status->set_code(kRegexpBadCharRange);
      status->set_error_arg(StringPiece(s->data(), 1+n));
      re->Decref();
      return false;
    }
    first = false;

    // Look for [:alnum:] etc.
    if (s->size() > 2 && (*s)[0] == '[' && (*s)[1] == ':') {
      switch (ParseCCName(s, flags_, re->ccb_, status)) {
        case kParseOk:
          continue;
        case kParseError:
          re->Decref();
          return false;
        case kParseNothing:
          break;
      }
    }

    // Look for Unicode character group like \p{Han}