Function ExtractAll

extensions/regex_ext.cc:89–145

}

Value ExtractAll(int regex_max_program_size, const StringValue& target,
                 const StringValue& regex,
                 const google::protobuf::DescriptorPool* absl_nonnull descriptor_pool,
                 google::protobuf::MessageFactory* absl_nonnull message_factory,
                 google::protobuf::Arena* absl_nonnull arena) {
  std::string target_scratch;
  std::string regex_scratch;
  absl::string_view target_view = target.ToStringView(&target_scratch);
  absl::string_view regex_view = regex.ToStringView(&regex_scratch);
  RE2 re2(regex_view, cel::internal::MakeRE2Options());
  CEL_RETURN_IF_ERROR(cel::internal::CheckRE2(re2, regex_max_program_size))
      .With(ErrorValueReturn());
  const int group_count = re2.NumberOfCapturingGroups();
  if (group_count > 1) {
    return ErrorValue(absl::InvalidArgumentError(absl::StrFormat(
        "regular expression has more than one capturing group: %s",
        regex_view)));
  }

  auto builder = NewListValueBuilder(arena);
  absl::string_view temp_target = target_view;

  // Space for the full match (\0) and the first capture group (\1).
  absl::string_view submatches[2];
  const int group_to_extract = (group_count == 1) ? 1 : 0;

  while (re2.Match(temp_target, 0, temp_target.length(), RE2::UNANCHORED,
                   submatches, group_count + 1)) {
    const absl::string_view& full_match = submatches[0];
    const absl::string_view& desired_capture = submatches[group_to_extract];

    // Avoid infinite loops on zero-length matches
    if (full_match.empty()) {
      if (temp_target.empty()) {
        break;
      }
      temp_target.remove_prefix(1);
      continue;
    }

    if (group_count == 1 && desired_capture.empty()) {
      temp_target.remove_prefix(full_match.data() - temp_target.data() +
                                full_match.length());
      continue;
    }

    absl::Status status =
        builder->Add(StringValue::From(desired_capture, arena));
    if (!status.ok()) {
      return ErrorValue(status);
    }
    temp_target.remove_prefix(full_match.data() - temp_target.data() +
                              full_match.length());
  }

  return std::move(*builder).Build();
}