Skip to content
Java client for Serialized
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
src
.gitignore
.travis.yml
README.md
pom.xml

README.md

Serialized Java Client

This is the official Java client for Serialized.

Build Status Maven Central

Getting started

Start by registering at https://serialized.io to get your access key and secret key. You will need them to configure the client.

You will create separate client instances for the different APIs (aggregates, feeds, projections and reactions).

Adding dependency

Add the following to your Maven POM file

<dependency>
  <groupId>io.serialized</groupId>
  <artifactId>serialized-client</artifactId>
  <version>1.2.3</version>
</dependency>

Configuring the client

Start by creating a configuration that will used for the client:

SerializedClientConfig serializedConfig = SerializedClientConfig.serializedConfig()
      .accessKey("<YOUR-ACCESS-KEY>")
      .secretAccessKey("<YOUR-SECRET-ACCESS-KEY>").build();

Creating an aggregate

To use the aggregate client you need two different classes.

The state implementation and the aggregate root implementation. The state is a mutable class that is used to materialize the current state from the events for a given aggregate type.

The aggregate root implementation is a Java class that contains a number of methods that handles commands. Each method in the aggregate root typically returns List<Event> that should contain 0..N events. These events will be saved atomically by the client during save/update.

You can choose how to organize the command handler methods, but it's recommended to keep them in a single class for each aggregate type.

In the following example the OrderState class contains the current state of the Order, which is the aggregate root:

public class OrderState {

  private OrderStatus status;
  private String orderId;

  public OrderState handleOrderPlaced(Event<OrderPlaced> event) {
    this.status = OrderStatus.PLACED;
    this.orderId = event.data().orderId;
    return this;
  }

  public OrderStatus status() {
    return status;
  }

  public String orderId() {
    return orderId;
  }
}
// Aggregate root Order
public class Order {

  private final OrderStatus status;
  private final String orderId;

  public Order(OrderState state) {
    this.status = state.status();
    this.orderId = state.orderId();
  }

  public List<Event> placeOrder(UUID orderId, long amount) {
    if (orderId.toString().equals(this.orderId)) {
      return emptyList();
    } else {
      return singletonList(orderPlaced(orderId.toString(), amount));
    }
  }

}

To create a client handling order events we create a client for the aggregates API by calling aggregateClient(String aggregateType, Class<T> stateClass, SerializedClientConfig config). We also register each event handler so that the client knows how to materialize the OrderState from the stream of events. This must be done for each event type.

AggregateClient<OrderState> orderClient = AggregateClient.aggregateClient("order", OrderState.class, serializedConfig)
    .registerHandler(OrderPlaced.class, OrderState::handleOrderPlaced)
    .build();

Saving events for an aggregate

When we want to save events for a new aggregate we start by initializing our aggregate root Order using a fresh and empty OrderState instance. The method placeOrder() contains our business logic and will return a list of events that we want to store. The client method save will automatically make sure no events are previously stored for our new aggregate (i.e. orderId). Under the hood this is achieved by including the field expectedVersion set to zero in the API call to Serialized.

OrderState orderState = new OrderState();
Order order = new Order(orderState);
List<Event> events = order.placeOrder(orderId, 123L);

orderClient.save(orderId, events);

Updating/appending events to an aggregate

When we want to update the aggregate, all previous events have to be loaded and our aggregate root Order has to be fast-forwarded to its current state. For convenience both the loading and the updating part is handled by the client method update.

orderClient.update(orderId, orderState -> {
  Order order = new Order(orderState);
  List<Event> events = order.cancel();
  return events;
});

The update method will automatically handle optimistic concurrency and make sure the aggregate is only updated if no other change has been performed to it. Under the hood this is achieved by including the field expectedVersion set to the same version number that was received when the aggregate was first loaded from Serialized.

Creating projections

To create projections using the client we need to create a ProjectionApiClient by calling projectionsClient(SerializedClientConfig config):

ProjectionApiClient projectionsClient = ProjectionApiClient.projectionsClient(serializedConfig).build();

There are numerous combinations of projection definitions that you can create and we encourage you to explore the documentation as well as the sample code.

To illustrate we can take the following example which will count the number of placed orders

projectionsClient.createOrUpdate(
        aggregatedProjection("order-count")
            .feed("order")
            .addHandler("OrderPlaced",
                inc().with(targetSelector("wins")).build())
            .build());

The createOrUpdate() uses the HTTP PUT API call which means that this method is is idempotent. A good practice is to place the code above in the initialization part of your service, which will automatically update the projection definition (and re-write all projection data from all events) if the definition code has changed since the last call.

Querying projections

The projection client also has methods for querying projections that you created. You can create a simple Java class that has the fields needed to deserialize the projection data and pass to the query methods of the projection client.

In this case we have a OrderCount class that looks like this:

public class OrderCount {
  public int totalPlacedOrders;
}

Querying our aggregated projection is now as simple as this:

Projection<OrderCount> projectionResponse = projectionApiClient.query(aggregated("order-count").build(OrderCount.class));
OrderCount theCount = projectionResponse.data;

Subscribing to event feeds

Start by creating a FeedClient like this

FeedClient feedClient = FeedClient.feedClient(getConfig()).build();

Point out which feed you want to subscribe to and start receiving events for all orders since the beginning of time.

FeedClient.FeedRequest request = feedClient.feed("order");
request.subscribe(feedEntry -> {
  // Process events in entry...
});
You can’t perform that action at this time.