Implement MongoDB Full Text Search with Spring Data MongoDB.

Background

MongoDB full text search provides the flexibility to perform search entries through multiple fields. In this example we will explore how to implement full text search with Spring Data MongoDB.

Verification

Given we have the following entries in Character:

Table 1. Characters
Name Publisher

Captain Marvel

Marvel

Joker

DC

Thanos

Marvel

When searching for captain marvel then the following results should be returned

Table 2. Characters that contains the keyword captain or marvel
Name Publisher

Captain Marvel

Marvel

Thanos

Marvel

This is demonstrated in CharacterRepositoryTests.

Implementation

We will start by defining the Character entity.

@Document
class Character {

    @Id
    private ObjectId id;

    @TextIndexed
    private final String name;

    @TextIndexed
    private final String publisher;

    public Character(String name, String publisher) {
        this.name = name;
        this.publisher = publisher;
    }

}

With Predefined Index

If respective fields are already indexed then we can utilise Spring Data query generation to perform full text search.

This can be done by creating a method that takes TextCriteria as parameter:

interface CharacterRepository extends MongoRepository<Character, ObjectId>, CharacterSearchRepository {

    List<Character> findAllBy(TextCriteria criteria, Sort sort);

}

This method can then be used in the following manner:

@Testcontainers
@DataMongoTest
class CharacterRepositoryTests {

    @Container
    @ServiceConnection
    private final static MongoDBContainer mongo = new MongoDBContainer("mongo:latest");

    @Autowired
    private MongoOperations operations;

    @Autowired
    private CharacterRepository repository;

    @Test
    @DisplayName("Generated query: Search for 'captain marvel' should return 'Captain Marvel' and 'Thanos'")
    void withGeneratedQuery() {
        // Simulate predefined index
        operations.indexOps(Character.class).ensureIndex(new TextIndexDefinitionBuilder().onFields("name", "publisher").build());

        var characters = repository.findAllBy(new TextCriteria().matchingAny("captain", "marvel"), Sort.by("name"));

        assertThat(characters)
                .hasSize(2)
                .extracting("name")
                .containsOnly("Captain Marvel", "Thanos")
                .doesNotContain("Joker");
    }

}

Without Predefined Index

Without predefined index, we will need to implement a custom repository implementation. We will start by defining a custom repository interface, CharacterSearchRepository:

interface CharacterSearchRepository {

    List<Character> findByText(String text, Sort sort);

}

Next, implement the custom repository interface in CharacterSearchRepositoryImpl:

class CharacterSearchRepositoryImpl implements CharacterSearchRepository {

    private final MongoOperations operations;

    CharacterSearchRepositoryImpl(MongoOperations operations) {
        this.operations = operations;
    }

    @Override
    public List<Character> findByText(String text, Sort sort) {
        operations.indexOps(Character.class)
                .ensureIndex(new TextIndexDefinitionBuilder().onFields("name", "publisher").build());

        var parameters = text.split(" ");
        var query = TextQuery.queryText(new TextCriteria().matchingAny(parameters)).with(sort);

        return operations.find(query, Character.class);
    }

}

This implementation will indexed searchable fields, i.e. name and publisher before searching the Document.

Finally, we will verify our custom implementation through integration test:

class CharacterRepositoryTests {

    @Container
    @ServiceConnection
    private final static MongoDBContainer mongo = new MongoDBContainer("mongo:latest");

    @Autowired
    private MongoOperations operations;

    @Autowired
    private CharacterRepository repository;

    void findByText() {
        var characters = repository.findByText("captain marvel", Sort.by("name"));

        assertThat(characters)
                .hasSize(2)
                .extracting("name")
                .containsOnly("Captain Marvel", "Thanos")
                .doesNotContain("Joker");
    }

}