Introduction to Spring Data Elasticsearch

1. Overview

In this tutorial, we'll explore the basics of Spring Data Elasticsearch in a code-focused and practical manner.

We'll learn how to index, search, and query Elasticsearch in a Spring application using Spring Data Elasticsearch. Spring Data Elasticseach is a Spring module that implements Spring Data, thus offering a way to interact with the popular open-source, Lucene-based search engine.

While Elasticsearch can work without a hardly defined schema, it's still a common practice to design one and create mappings specifying the type of data we expect in certain fields. When a document is indexed, its fields are processed according to their types. For example, a text field will be tokenized and filtered according to mapping rules. We can also create filters and tokenizers of our own.

For the sake of simplicity, we'll use a docker image for our Elasticsearch instance, though any Elasticsearch instance listening on port 9200 will do.

We'll start by firing up our Elasticsearch instance:

docker run -d --name es762 -p 9200:9200 -e "discovery.type=single-node" elasticsearch:7.6.2

2. Spring Data

Spring Data helps avoid boilerplate code. For example, if we define a repository interface that extends the ElasticsearchRepository interface that Spring Data Elasticsearch provides, CRUD operations for the corresponding document class will become available by default.

Additionally, method implementations will generate for us simply by declaring methods with names in a predefined format. There's no need to write an implementation of the repository interface.

The Baeldung guides on Spring Data provide the essentials to get started on this topic.

2.1. Maven Dependency 

Spring Data Elasticsearch provides a Java API for the search engine. In order to use it, we need to add a new dependency to the pom.xml:

<dependency>
    <groupId>org.springframework.data</groupId>
    <artifactId>spring-data-elasticsearch</artifactId>
    <version>4.0.0.RELEASE</version>
</dependency>

2.2. Defining Repository Interfaces 

In order to define new repositories, we'll extend one of the provided repository interfaces, replacing the generic types with our actual document and primary key types.

It's important to note that ElasticsearchRepository extends from PagingAndSortingRepository. This allows built-in support for pagination and sorting.

In our example, we'll use the paging feature in our custom search methods:

public interface ArticleRepository extends ElasticsearchRepository<Article, String> {

    Page<Article> findByAuthorsName(String name, Pageable pageable);

    @Query("{\"bool\": {\"must\": [{\"match\": {\"authors.name\": \"?0\"}}]}}")
    Page<Article> findByAuthorsNameUsingCustomQuery(String name, Pageable pageable);
}

With the findByAuthorsName method, the repository proxy will create an implementation based on the method name. The resolution algorithm will determine that it needs to access the authors property, and then search the name property of each item.

The second method, findByAuthorsNameUsingCustomQuery, uses a custom Elasticsearch boolean query defined using the @Query annotation, which requires strict matching between the author's name and the provided name argument.

2.3. Java Configuration

When configuring Elasticsearch in our Java application, we need to define how we connect to the Elasticsearch instance. For that, we'll use a RestHighLevelClient, which the Elasticsearch dependency offers:

@Configuration
@EnableElasticsearchRepositories(basePackages = "com.baeldung.spring.data.es.repository")
@ComponentScan(basePackages = { "com.baeldung.spring.data.es.service" })
public class Config {

    @Bean
    public RestHighLevelClient client() {
        ClientConfiguration clientConfiguration 
            = ClientConfiguration.builder()
                .connectedTo("localhost:9200")
                .build();

        return RestClients.create(clientConfiguration).rest();
    }

    @Bean
    public ElasticsearchOperations elasticsearchTemplate() {
        return new ElasticsearchRestTemplate(client());
    }
}

We're using a standard Spring-enabled style annotation. @EnableElasticsearchRepositories will make Spring Data Elasticsearch scan the provided package for Spring Data repositories.

In order to communicate with our Elasticsearch server, we'll use a simple RestHighLevelClient. While Elasticsearch provides multiple types of clients, using the RestHighLevelClient is a good way to future-proof the communication with the server.

Finally, we'll set up an ElasticsearchOperations bean to execute operations on our server. In this case, we instantiate an ElasticsearchRestTemplate.

3. Mappings

We use mappings to define a schema for our documents. By defining a schema for our documents, we protect them from undesired outcomes, such as mapping to an unwanted type.

Our entity is a simple document, Article, where the id is of the type String. We'll also specify that such documents must be stored in an index named blog within the article type.

@Document(indexName = "blog", type = "article")
public class Article {

    @Id
    private String id;
    
    private String title;
    
    @Field(type = FieldType.Nested, includeInParent = true)
    private List<Author> authors;
    
    // standard getters and setters
}

Indexes can have several types, which we can use to implement hierarchies.

We'll mark the authors field as FieldType.Nested. This allows us to define the Author class separately, but still have the individual instances of author embedded in an Article document when it's indexed in Elasticsearch.

4. Indexing Documents 

Spring Data Elasticsearch generally auto-creates indexes based on the entities in the project. However, we can also create an index programmatically via the client template:

elasticsearchTemplate.indexOps(Article.class).create();

Then we can add documents to the index:

Article article = new Article("Spring Data Elasticsearch");
article.setAuthors(asList(new Author("John Smith"), new Author("John Doe")));
articleRepository.save(article);

5. Querying

5.1. Method Name-Based Query

When we use the method name-based query, we write methods that define the query we want to perform. During the setup, Spring Data will parse the method signature and create the queries accordingly:

String nameToFind = "John Smith";
Page<Article> articleByAuthorName
  = articleRepository.findByAuthorsName(nameToFind, PageRequest.of(0, 10));

By calling findByAuthorsName with a PageRequest object, we'll obtain the first page of results (page numbering is zero-based), with that page containing at most 10 articles. The page object also provides the total number of hits for the query, along with other handy pagination information.

5.2. A Custom Query

There are a couple of ways to define custom queries for Spring Data Elasticsearch repositories. One way is to use the @Query annotation, as demonstrated in section 2.2.

Another option is to use the query builder to create our custom query.

If we want to search for articles that have the word “data” in the title, we can just create a NativeSearchQueryBuilder with a Filter on the title:

Query searchQuery = new NativeSearchQueryBuilder()
   .withFilter(regexpQuery("title", ".*data.*"))
   .build();
SearchHits<Article> articles = 
   elasticsearchTemplate.search(searchQuery, Article.class, IndexCoordinates.of("blog");

6. Updating and Deleting 

In order to update a document, we must first retrieve it:

String articleTitle = "Spring Data Elasticsearch";
Query searchQuery = new NativeSearchQueryBuilder()
  .withQuery(matchQuery("title", articleTitle).minimumShouldMatch("75%"))
  .build();

SearchHits<Article> articles = 
   elasticsearchTemplate.search(searchQuery, Article.class, IndexCoordinates.of("blog");
Article article = articles.getSearchHit(0).getContent();

Then we can make changes to the document by editing the content of the object using its assessors:

article.setTitle("Getting started with Search Engines");
articleRepository.save(article);

As for deleting, there are several options. We can retrieve the document and delete it using the delete method:

articleRepository.delete(article);

We can also delete it by id once we know it:

articleRepository.deleteById("article_id");

It's also possible to create custom deleteBy queries and make use of the bulk delete feature offered by Elasticsearch:

articleRepository.deleteByTitle("title");

7. Conclusion

In this article, we explored how to connect and make use of Spring Data Elasticsearch. We discussed how to query, update, and delete documents. Finally, we learned how to create custom queries if what's offered by Spring Data Elasticsearch doesn't fit our needs.

As usual, the source code used throughout this article can be found over on GitHub.