I have a use case in which I want to index my document in a separate index based on a certain condition.
For example, I want to store an invoice document to an index with department name suffixed with it.
@Document(indexName="store_{department}", indexStoreType="invoice")
public class InvoiceES{
// fields
@Id
private String id;
@Field
private String department;
}
Is it possible to achieve this using Spring Data?
If not, is it planned in a coming release of Spring Data?
As far as the spring-boot-starter-data-elasticsearch-1.5, you can achieve it by spring el expression:
@Bean
Department department() {
return new Department();
}
@Document(indexName="store_#{department.name()}", indexStoreType="invoice")
public class InvoiceES{}
You can change the bean's property to change the index you want to save/search:
invoiceRepo.save(new Invoice(...));
department.setName("newName");
invoiceRepo.save(new Invoice(...));
What should be noticed is not to share this bean in multiple thread, which may mess up your index.
I had to solve similar problem to store data with dynamic index name, consisting of current date: "Index-name-YYYY.MM.DD".
I made a component which has getter of formatted date string:
@Component
public class ElasticIndex {
private static final DateTimeFormatter formatter = DateTimeFormatter.ofPattern("yyyy.MM.dd");
public String getIndexDate() {
return formatter.format(LocalDate.now());
}
}
Thus, your document class will be like that:
@Document(indexName = "my_data-#{elasticIndex.getIndexDate()}", type = "DataDoc")
public class DataDoc {
@Id
private String id;
private String dataToStore;
....
}
So each time you save/delete/etc new document to your elasticsearch repository, it gets index name from getIndexDate method of ElasticIndex bean.
Works on Spring Boot 2.2.5, Spring Data Elasticsearch 3.2.5
@Document(indexName = "#{'${elasticsearch.index.name}'}", type = "category", shards = 1, replicas = 0, refreshInterval = "-1")
It created index: "${elasticsearch.index.name}"
I tried with spring-data-elasticsearch version 1.1.2.RELEASE and 1.2.0.M1 but according to jira (https://jira.spring.io/browse/DATAES-93) it is fixed in version 1.1 RC1
Vishal,
Currently spring data elasticsearch does not support this feature. we already have feature request(pull request) which will be added soon with next release.
Have a look on this pull request, https://github.com/spring-projects/spring-data-elasticsearch/pull/56
Spring will create Elasticsearch[Rest]Template bean which implements ElasticsearchOperations.
The ElasticsearchOperations interface offer index(IndexQuery query) function to save a document.
You can use IndexQueryBuilder.withIndexName to overwrite the index name.
Based on the pull request link posted earlier, I was able to stitch together a solution. Currently, support is there only for beans of type String, and they can be injected into the attribute using Spring Expression Language (SpEL), with special @ notation.
First Define a Bean that returns the dynamic prefix:
@Bean
String department() {
return "some-department";
}
Next, use that in the Elastic Document. Please make a note of the special @ notation.
@Document(indexName = "#{@department}_department")
Notes: Actual requirement for Spring.
The only way I found to do this is doing it manually without @Document anotation:
client = new TransportClient().addTransportAddress(new InetSocketTransportAddress(nodeId, port));
IndexResponse response = client.prepareIndex(your_index, type, subid)
.setSource(jsonBuilder()
.startObject()
.field("field1", field1))
.field("fileld2", field2)
).endObject())
.execute().actionGet();
indexNamein this case) in code is not a good idea, not sure why Spring folks took this decision.