Function ValidateRegex

test/gtest/gtest-all.cc:8467–8521

Generates non-fatal failures and returns false if regex is invalid; otherwise returns true.

// Generates non-fatal failures and returns false if regex is invalid;
// otherwise returns true.
bool ValidateRegex(const char* regex) {
  if (regex == NULL) {
    // TODO(wan@google.com): fix the source file location in the
    // assertion failures to match where the regex is used in user
    // code.
    ADD_FAILURE() << "NULL is not a valid simple regular expression.";
    return false;
  }

  bool is_valid = true;

  // True iff ?, *, or + can follow the previous atom.
  bool prev_repeatable = false;
  for (int i = 0; regex[i]; i++) {
    if (regex[i] == '\\') {  // An escape sequence
      i++;
      if (regex[i] == '\0') {
        ADD_FAILURE() << FormatRegexSyntaxError(regex, i - 1)
                      << "'\\' cannot appear at the end.";
        return false;
      }

      if (!IsValidEscape(regex[i])) {
        ADD_FAILURE() << FormatRegexSyntaxError(regex, i - 1)
                      << "invalid escape sequence \"\\" << regex[i] << "\".";
        is_valid = false;
      }
      prev_repeatable = true;
    } else {  // Not an escape sequence.
      const char ch = regex[i];

      if (ch == '^' && i > 0) {
        ADD_FAILURE() << FormatRegexSyntaxError(regex, i)
                      << "'^' can only appear at the beginning.";
        is_valid = false;
      } else if (ch == '$' && regex[i + 1] != '\0') {
        ADD_FAILURE() << FormatRegexSyntaxError(regex, i)
                      << "'$' can only appear at the end.";
        is_valid = false;
      } else if (IsInSet(ch, "()[]{}|")) {
        ADD_FAILURE() << FormatRegexSyntaxError(regex, i)
                      << "'" << ch << "' is unsupported.";
        is_valid = false;
      } else if (IsRepeat(ch) && !prev_repeatable) {
        ADD_FAILURE() << FormatRegexSyntaxError(regex, i)
                      << "'" << ch << "' can only follow a repeatable token.";
        is_valid = false;
      }

      prev_repeatable = !IsInSet(ch, "^$?*+");
    }
  }

  return is_valid;
}

// Matches a repeated regex atom followed by a valid simple regular
// expression.  The regex atom is defined as c if escaped is false,