Method ParseKey

cpp/src/arrow/dataset/partition.cc:689–727

}

Result<std::optional<KeyValuePartitioning::Key>> HivePartitioning::ParseKey(
    const std::string& segment, const HivePartitioningOptions& options) {
  auto name_end = string_view(segment).find_first_of('=');
  // Not round-trippable
  if (name_end == string_view::npos) {
    return std::nullopt;
  }

  // Static method, so we have no better place for it
  util::InitializeUTF8();

  std::string name;
  std::string value;
  switch (options.segment_encoding) {
    case SegmentEncoding::None: {
      name = segment.substr(0, name_end);
      value = segment.substr(name_end + 1);
      if (ARROW_PREDICT_FALSE(!util::ValidateUTF8(segment))) {
        return Status::Invalid("Partition segment was not valid UTF-8: ", segment);
      }
      break;
    }
    case SegmentEncoding::Uri: {
      auto raw_value = std::string_view(segment).substr(name_end + 1);
      ARROW_ASSIGN_OR_RAISE(value, SafeUriUnescape(raw_value));
      auto raw_key = std::string_view(segment).substr(0, name_end);
      ARROW_ASSIGN_OR_RAISE(name, SafeUriUnescape(raw_key));
      break;
    }
    default:
      return Status::NotImplemented("Unknown segment encoding: ",
                                    options.segment_encoding);
  }

  if (value == options.null_fallback) {
    return Key{std::move(name), std::nullopt};
  }
  return Key{std::move(name), std::move(value)};
}

Result<std::vector<KeyValuePartitioning::Key>> HivePartitioning::ParseKeys(
    const std::string& path) const {