Spring Data Tarantool

The primary goal of the Spring Data project is to make it easier to build Spring-powered applications that use new data access technologies such as non-relational databases, map-reduce frameworks, and cloud based data services.

The Spring Data Tarantool project provides Spring Data way of working with Tarantool database spaces. The key features are repository support and different API layers which are common in Spring Data, as well as support for asynchronous approaches baked by the underlying asynchronous Tarantool database driver.

Spring Boot compatibility

Tarantool compatibility

References

The Tarantool Database documentation is located at tarantool.io

Feel free to join the Tarantool community chat in Telegram (or its counterpart in Russian) if you have any questions about Tarantool database or Spring Data Tarantool.

Detailed questions can be asked on StackOverflow using the tarantool tag.

Documentation and StackOverflow links will be added in the nearest future.

If you are new to Spring as well as to Spring Data, look for information about Spring projects.

Quick Start

Demo project

Check out a traditional Pet Clinic application implemented using this module.

Maven configuration

Add the Maven dependency:

<dependency>
    <groupId>io.tarantool</groupId>
    <artifactId>spring-data-tarantool</artifactId>
    <version>0.6.1</version>
</dependency>

TarantoolTemplate

TarantoolTemplate is the central support class for Tarantool database operations. It provides:

• Basic POJO mapping support to and from Tarantool tuples
• Convenience methods to interact with the store (insert object, update objects, select)
• Exception translation into Spring's technology agnostic DAO exception hierarchy.

Spring Data repositories

To simplify the creation of data repositories Spring Data Tarantool provides a generic repository programming model. It will automatically create a repository proxy for you that adds implementations of finder methods you specify on an interface.

Only simple CRUD operations including entity ID are supported at the moment (will be fixed soon).

For example, given a Book class with name and author properties, a BookRepository interface can be defined for saving and loading the entities:

@Tuple("books")
public class Book {
    @Id
    private Integer id;

    @Field(name = "unique_key")
    private String uniqueKey;

    @Field(name = "book_name")
    private String name;

    private String author;

    private Integer year;
}

public interface BookRepository extends CrudRepository<Book, Long> {
}

The @Tuple annotation allows to specify the space name which schema will be used for forming the data tuples to sending and for mapping the incoming tuple into an object. You can omit the @Tuple annotation, then the connector will take the class name converted to snake_case. The @Field annotations provide custom names for the fields. It is necessary to have at least one field marked with the @Id annotation.

Extending CrudRepository causes basic CRUD methods being pulled into the interface so that you can easily save and find single entities and collections of them. We can also use Tarantool specific interface TarantoolRepository.

Let's assume that you have the tarantool/crud module installed in yor Cartridge application and there is an active 'crud-router' role on your router. Put the following settings in your application properties:

Than you can have Spring automatically create a proxy for the repository interface by using the following JavaConfig:

@Configuration
@EnableTarantoolRepositories(basePackageClasses = BookRepository.class)
class ApplicationConfig extends AbstractTarantoolDataConfiguration {

    @Value("${tarantool.host}")
    protected String host;
    @Value("${tarantool.port}")
    protected int port;
    @Value("${tarantool.username}")
    protected String username;
    @Value("${tarantool.password}")
    protected String password;

    @Override
    protected TarantoolServerAddress tarantoolServerAddress() {
        return new TarantoolServerAddress(host, port);
    }

    @Override
    public TarantoolCredentials tarantoolCredentials() {
        return new SimpleTarantoolCredentials(username, password);
    }

    @Override
    public TarantoolClient tarantoolClient(TarantoolClientConfig tarantoolClientConfig,
                                           TarantoolClusterAddressProvider tarantoolClusterAddressProvider) {
        return new ProxyTarantoolTupleClient(super.tarantoolClient(tarantoolClientConfig, tarantoolClusterAddressProvider));
    }
}

This sets up a connection to a local Tarantool instance and enables the detection of Spring Data repositories (through @EnableTarantoolRepositories).

This will find the repository interface and register a proxy object in the container. You can use it as shown below:

@Service
public class MyService {

    private final BookRepository repository;

    @Autowired
    public MyService(BookRepository repository) {
        this.repository = repository;
    }

    public void doWork() {
        Book book = new Book();
        book.setName("Le Petit Prince");
        book.setAuthor("Antoine de Saint-Exupéry");
        Book savedBook = repository.save(book);

        List<Book> allBooks = repository.findAll();
    }
}

Proxy Tarantool functions in repositories

Consider we need to write a complex query in Lua, working with sharded data in Tarantool Cartridge. In this case we can expose this query as a public API function and map that function on a repository method via the @Query annotation by specifying the functionName parameter:

public interface BookRepository extends CrudRepository<Book, Long> {
    @Query(function = "find_by_complex_query")
    List<Book> findByYearGreaterThenProxy(Integer year);
}

The corresponding function in on Tarantool Cartridge router may look like (uses the tarantool/crud module):

    local crud = require('crud')
    local fun = require('fun')

    ...

    function find_by_complex_query(year)
        return crud.pairs('books'):filter(function(b) return b[6] and b[6] > year end):totable()
    end

See more examples in the module tests.

Specify output

For such methods, you can specify the stored function response format so that it will be parsed correctly. The response format can be specified in the @Query annotation using the output parameter, see the examples below.

public interface SampleUserRepository extends TarantoolRepository<SampleUser, String> {
    @Query(function = "get_users_with_age_gt", output = TarantoolSerializationType.TUPLE)
    List<SampleUser> usersWithAgeGreaterThen(Integer age);

    @Query(function = "get_predefined_user", output = TarantoolSerializationType.AUTO)
    SampleUser predefinedUser();
}

function get_users_with_gt_age(age)
    return crud.select("sample_user", { { ">", "age", age } })
end

function get_predefined_user()
    return { name = "John", age = 46 }
end

The first function returns users who are older than the specified age. In the Repository, we indicate that we expect TUPLE, because crud.select returns the result in compressed format, flatten tuples. The keys for mapping the result into a Java object are obtained from the space metadata determined by the class name in the method return type. If this class has a @Tuple annotation with a custom space name, this space will be used for getting the metadata.

The second function returns a ready-made result with keys. So TUPLE will not work here, because it's not flatten structure, and we specify AUTO. Mapping happens by keys of table.

Of course, the word AUTO means that we can accept any result, and therefore it can be specified in the first request as well. But we specify TUPLE because a different stack of converters is used and the conversion is faster. For ease of use and clarity, the default output parameter is TarantoolSerializationType.AUTO.

We can also return primitive types if needed:

@Query(function = "get_age_by_name", output = TarantoolSerializationType.AUTO)
Optional<Integer> getAgeByName(String name);

function get_age_by_name(name)
    local user = crud.get("sample_user", name)
    if user.rows[1] ~= nil then
        return user.rows[1][2]
    end
    return nil
end

Composite primary key

You can create an entity representing a Tarantool tuple with composite primary index. For this you need to use @TarantoolIdClass annotation on entity to specify the type of id. Also you may mark all 'id' fields in the entity with standard @Id annotation. @Id annotation on properties is optional but It is recommended to use it to make code more clear.

See the example:

public class BookTranslationId {
    private Integer bookId;
    private String language;
    private Integer edition;
}

@Tuple("book_translation")
@TarantoolIdClass(BookTranslationId.class)
public class BookTranslation {
    @Id
    @Field(value = "id")
    private Integer bookId;
    @Id
    private String language;
    @Id
    private Integer edition;
    private String translator;
    private String comments;
}

public interface BookTranslationRepository
        extends TarantoolRepository<BookTranslation, BookTranslationId> {
}

Contributing to Spring Data Tarantool

Contributions and issues are welcome, feel free to add them to this project or offer directly in the Tarantool community chat or on StackOverflow using the tarantool tag.