Skip to content

Latest commit

 

History

History
72 lines (53 loc) · 2.72 KB

README.md

File metadata and controls

72 lines (53 loc) · 2.72 KB
Raw

Readme

This example uses an embedded Undertow web server and produces a fat Jar containing all needed dependencies.

Sources for the framework used can be found here: github - neo4j-websockets

Jar files for the framework are available from Maven Central Repository: maven.org - neo4j-websockets

Setup for Your Application

Configuration and Startup

This framework is a plugin for the server side. Hence, when using the provided example pom for maven, a fat Jar will be created that can be copied into the plugins directory of a Neo4j (2.3) server (community or enterprise).

To be able to pickup the functionality added by yourself, you must let the framework know which packages need to be scanned for Spring beans. To do this, simply add a line like the following to conf/neo4j.properties

websocket_packages=de.oliverwetterau.neo4j.examples.websockets.server

Incoming Messages

All incoming messages should to be replied with a Result so that can easily do some error handling on the client side.

@MessageMethod
public Result<TranslatedMovie> getByYear(JsonNode jsonNode) {
    return moviesService.getByYear(jsonNode.get("year").asInt());
}

The websocket framework provides a @Transactional annotation which creates a new Neo4j transaction if none is active and closes that transaction after the message call. Any uncaught exceptions in method surrounded with that annotation will be converted into an error that will be put into the Result (provided that method returns a Result object).

@Transactional
public Result<TranslatedMovie> getByYear(Integer year) {
    Result<TranslatedMovie> movieResult = new Result<>();

    graphDatabaseService.findNodes(Movie.MOVIE, Movie.YEAR, year)
            .forEachRemaining(node -> movieResult.add(TranslatedMovie.readMovie(node, exampleThreadLocale.getLocale())));

    return movieResult;
}

To use that functionality the application needs Aspectj compile time weaving as shown in the pom.xml.

If you do not want to use @Transactional you can use plain Neo4j transaction management:

public Result<Actor> create(String lastname, String firstname) {
    Result<Actor> actorResult = new Result<>();

    try (Transaction tx = neo4j.getDatabase().beginTx()) {
        actorResult.add(new Actor(graphDatabaseService.createNode(), lastname, firstname));
        actorResult.close();    // generate final json string before end of transaction

        tx.success();
    }
    catch (Exception e) {
        actorResult.add(exceptionToErrorConverter.convert(e));
    }

    return actorResult;
}