Fake implementation of the core MongoDB server in Java that can be used for integration tests.
Think of H2/HSQLDB/SQLite but for MongoDB.
The MongoDB Wire Protocol is implemented with Netty. Different backends are possible and can be extended.
In-Memory backend
The in-memory backend is the default backend that is typically used to fake MongoDB for integration tests. It supports most CRUD operations, commands and the aggregation framework. Some features are not yet implemented, such as full-text search or map/reduce.
Add the following Maven dependency to your project:
<dependency> <groupId>de.bwaldvogel</groupId> <artifactId>mongo-java-server</artifactId> <version>1.9.8</version> </dependency>
Example
public class SimpleTest { private MongoCollection<Document> collection; private MongoClient client; private MongoServer server; @Before public void setUp() { server = new MongoServer(new MemoryBackend()); // bind on a random local port InetSocketAddress serverAddress = server.bind(); client = new MongoClient(new ServerAddress(serverAddress)); collection = client.getDatabase("testdb").getCollection("testcollection"); } @After public void tearDown() { client.close(); server.shutdown(); } @Test public void testSimpleInsertQuery() throws Exception { assertEquals(0, collection.count()); // creates the database and collection in memory and insert the object Document obj = new Document("_id", 1).append("key", "value"); collection.insertOne(obj); assertEquals(1, collection.count()); assertEquals(obj, collection.find().first()); } }
Example with SpringBoot
@RunWith(SpringRunner.class) @SpringBootTest(classes={SimpleSpringBootTest.TestConfiguration.class}) public class SimpleSpringBootTest { @Autowired private MyRepository repository; @Before public void setUp() { // initialize your repository with some test data repository.deleteAll(); repository.save(...); } @Test public void testMyRepository() { // test your repository ... ... } @Configuration @EnableMongoRepositories(basePackageClasses={MyRepository.class}) protected static class TestConfiguration { @Bean public MongoTemplate mongoTemplate(MongoClient mongoClient) { return new MongoTemplate(mongoDbFactory(mongoClient)); } @Bean public MongoDbFactory mongoDbFactory(MongoClient mongoClient) { return new SimpleMongoDbFactory(mongoClient, "test"); } @Bean(destroyMethod="shutdown") public MongoServer mongoServer() { MongoServer mongoServer = new MongoServer(new MemoryBackend()); mongoServer.bind(); return mongoServer; } @Bean(destroyMethod="close") public MongoClient mongoClient(MongoServer mongoServer) { return new MongoClient(new ServerAddress(mongoServer.getLocalAddress())); } } }
H2 MVStore backend
The H2 MVStore backend connects the server to a MVStore that
can either be in-memory or on-disk.
<dependency> <groupId>de.bwaldvogel</groupId> <artifactId>mongo-java-server-h2-backend</artifactId> <version>1.9.8</version> </dependency>
Example
public class Application { public static void main(String[] args) throws Exception { MongoServer server = new MongoServer(new H2Backend("database.mv")); server.bind("localhost", 27017); } }
PostgreSQL backend
The PostgreSQL backend is a proof-of-concept implementation that connects the server to a database in a running PostgreSQL 9.5+ instance. Each MongoDB database is mapped to a schema in Postgres and each MongoDB collection is stored as a table.
<dependency> <groupId>de.bwaldvogel</groupId> <artifactId>mongo-java-server-postgresql-backend</artifactId> <version>1.9.8</version> </dependency>
For integration tests, a PostgreSQL instance can be created in a docker container:
docker run --name postgres-mongo-java-server-test -p 5432:5432 -e POSTGRES_USER=mongo-java-server-test -e POSTGRES_PASSWORD=mongo-java-server-test -e POSTGRES_DB=mongo-java-server-test -d postgres:9.6-alpine
Example
public class Application { public static void main(String[] args) throws Exception { DataSource dataSource = new org.postgresql.jdbc3.Jdbc3PoolingDataSource(); dataSource.setDatabaseName(…); dataSource.setUser(…); dataSource.setPassword(…); MongoServer server = new MongoServer(new PostgresqlBackend(dataSource)); server.bind("localhost", 27017); } }
Ideas for other backends
Faulty backend
A faulty backend could randomly fail queries or cause timeouts. This could be used to test the client for error resilience.
Fuzzy backend
Fuzzing the wire protocol could be used to check the robustness of client drivers.
Related Work
-
- Spins up a real MongoDB instance
-
- focus on unit testing
- no wire protocol implementation
- intercepts the java mongo driver
- currently used in nosql-unit
