Repository Context with Convertors

Convertor Context Passing

Context data flows naturally through Convertor input parameters and subscription context.

Repository Context with Convertors

Overview

In the Convertor pattern, context data (user IDs, chat IDs, filters) flows naturally through typed input parameters. There is no separate “RequestContext” object - the Convertor’s input parameter carries all needed context.

Passing Context Through Parameters

Filter Context

Stream<List<NotificationModel>> queryList({
  required String eventId,
  NotificationFilterCriteria? additionalFilter,
}) {
  final filter = NotificationFilterCriteria(
    systemParentId: eventId, // Context: current event
    ...additionalFilter,
  );

  final params = ListRequestParams<GnotificationFilters, GnotificationOrder>(
    filter: filterConvertor.mightExecute(filter),
  );

  return listDatasource.thenEach(modelConvertor).execute(params);
}

Subscription Context

For subscriptions that need filtering context (e.g., chat messages for a specific chat):

Stream<ChatMessageModel> subscribeToMessages({
  required String chatId,
  required String currentUserId,
}) {
  final context = ChatMessageSubscriptionContext(
    chatId: chatId,
    currentUserId: currentUserId,
  );

  final params = SubscriptionRequestParams(context: context);
  return subscriptionDatasource.thenMap(modelConvertor).execute(params);
}

The subscription context is serialized to JSON and passed via Ferry’s updateCacheHandlerContext:

@riverpod
Convertor<GSubscribeToChatMessagesReq,
    SubscriptionRequestParams<ChatMessageSubscriptionContext>>
chatMessageSubscriptionConvertor(Ref ref) {
  return Convertor((params) {
    return GSubscribeToChatMessagesReq((builder) {
      builder
        ..requestId = 'chat_message_subscription'
        ..updateCacheHandlerKey = 'messageSubscriptionCacheHandler'
        ..updateCacheHandlerContext = params.context?.toJson();
    });
  });
}

Using Interceptors for Cross-Cutting Context

Add logging or analytics without modifying the data flow:

final loggedDatasource = datasource.intercept(
  onInput: (params) => log('Fetching with context: ${params.filter}'),
  onOutput: (data) => log('Received ${data.length} items'),
);

Using Decorators for Conditional Logic

Apply context-dependent behavior:

final withErrorHandling = datasource.decorate((convertor, input) {
  try {
    return convertor.execute(input);
  } catch (e, st) {
    analytics.trackError('datasource_error', error: e);
    return Stream.error(DomainException('Failed to fetch data'), st);
  }
});