Hibernate Search is a library that integrates Hibernate ORM with Apache Lucene or Elasticsearch by automatically indexing entities, enabling advanced search functionality: full-text, geospatial, aggregations and more. For more information, see Hibernate Search on hibernate.org.

We just published Hibernate Search 6.0.0.Alpha6, a new release of the still-in-development 6.0 branch. This release mainly restores decent performance for Lucene index reads and writes, adds support for indexing BigDecimal and BigInteger, reduces the verbosity of the search DSL, and adds support for configuring on-commit synchronization for automatic indexing.

Getting started with Hibernate Search 6

If you want to dive right into the new, shiny Hibernate Search 6, a good starting point is the getting started guide included in the reference documentation.

Hibernate Search 6 is still in development and its APIs differ significantly from Search 5.

For more information about the current status of this branch, see the page dedicated to Search 6 on hibernate.org.

For more information about migration and what we intend to do to help you, see the migration guide.

What’s new

Orchestration of Lucene index accesses

The Lucene backend now more efficiently orchestrates index writes and reads, delegating those to background threads like it used to in Search 5.

Though no benchmarks were attempted yet, performance should be back to a level comparable to Search 5.

Indexing of BigDecimal and BigInteger as scaled numbers

BigDecimal and BigInteger can now be indexed with Hibernate Search out of the box.

For predicates (match, range, …​) and sorts, these numbers will be scaled down and indexed as a long, keeping only the most significant digits.

Projections are not affected by scaling: projections on a BigDecimal or BigInteger will return the original number, not just the most significant digits.

By default, the scale used will be extracted from Hibernate ORM’s @Column.scale attribute. You can override that with Hibernate Search’s @ScaledNumberField annotation and its decimalScale attribute.

For more information, see the section of the documentation about @ScaledNumberField.

Search API improvements

The Search API has once again been improved, mainly by reducing its verbosity:

  • .asEntity() can now be omitted (HSEARCH-3578);

  • .toQuery() can now be omitted (HSEARCH-3572);

  • extensions can now be used at the very start of the DSL (HSEARCH-3544), and will be transparently applied to all steps: projections, predicates, sorts.

As a result, this:

SearchResult<Book> result = Search.getSearchSession(em).search(Book.class)
        .asEntity()
        .predicate(f -> f.extension(ElasticsearchExtension.get()).fromJson("{'match_all': {}}"))
        .sort(f -> f.extension(ElasticsearchExtension.get()).fromJson("{'field1': 'asc'}"))
        .toQuery()
        .fetch(20, 0);

Can now be written like this:

SearchResult<Book> result = Search.getSearchSession(em).search(Book.class)
        .extension(ElasticsearchExtension.get())
        .predicate(f -> f.fromJson("{'match_all': {}}"))
        .sort(f -> f.fromJson("{'field1': 'asc'}"))
        .fetch(20, 0);

Mapping API improvements

As of HSEARCH-3463, the syntax for specifying container extractors in the annotation mapping is now less error-prone and more future-proof.

In short, this:

// Default: automatically resolved
@GenericField
private List<Integer> field1;
// Explicit extractors (built-in)
@GenericField(extractors = @ContainerExtractorRef(BuiltinContainerExtractor.MAP_KEY))
private Map<Integer, String> field2;
// No extractor at all
@GenericField(extractors = {}, valueBridge = @ValueBridgeRef(type = MyOptionalIntBridge.class))
private OptionalInt field3;
// Explicit extractors (builtin, multiple)
@GenericField(extractors = {
        @ContainerExtractorRef(BuiltinContainerExtractor.MAP_KEY),
        @ContainerExtractorRef(BuiltinContainerExtractor.COLLECTION),
}))
private Map<List<Integer>, String> field4;
// Explicit extractors (custom)
@GenericField(extractors = @ContainerExtraction({
        @ContainerExtractorRef(type = MyContainerExtractor.class),
        @ContainerExtractorRef(BuiltinContainerExtractor.COLLECTION)
}))
private MyContainer<List<Integer>> field5;

Should now be written like this:

// Default: automatically resolved
@GenericField
private List<Integer> field1;
// Explicit extractors (built-in)
@GenericField(extraction = @ContainerExtraction(BuiltinContainerExtractors.MAP_KEY))
private Map<Integer, String> field2;
// No extractor at all
@GenericField(extraction = @ContainerExtraction(extract = ContainerExtract.NO))
private OptionalInt field3;
// Explicit extractors (builtin, multiple)
@GenericField(extraction = @ContainerExtraction({
        BuiltinContainerExtractors.MAP_KEY,
        BuiltinContainerExtractors.COLLECTION,
}))
private Map<List<Integer>, String> field4;
// Explicit extractors (custom) => Requires the names to be assigned to classes using a mapping configurer
@GenericField(extraction = @ContainerExtraction({
        "myExtractor1",
        BuiltinContainerExtractors.COLLECTION
}))
private MyContainer<List<Integer>> field5;

Configurable synchronization for automatic indexing

In HSEARCH-3316, we reworked how to configure the behavior of automatic indexing, in particular how much Hibernate Search should wait for indexing to finish on transaction commits.

You can now set the configuration property hibernate.search.automatic_indexing.synchronization_strategy to one of three values:

  • queued to simply queue the indexing tasks in memory and not wait at all;

  • committed (the default) to wait for indexing tasks to be processed and committed to persistent storage before returning from the database commit;

  • searchable to do the same as committed, and then also wait for changes to be visible in indexes, so that a search query executed just after the database commit will be consistent with the committed transaction, without any further delay.

Each synchronization strategy has pros and cons. Depending on your application, you may opt for queued to favor throughput or searchable to ensure the index is always immediately up-to-date after a transaction is committed. The last one is particularly suitable for automated tests.

Note that configuration goes beyond just this configuration property: you can override the setting on a per-session basis (to temporarily set it to queued when performing batch operations, for example) or even define your own, custom strategy.

For more information, head to the section of the documentation about automatic indexing.

Search hit explanation

As of HSEARCH-3353, Lucene and Elasticsearch queries now once again expose explain() methods, providing a description of the computation that lead to the score of a given document.

Use .extension(LuceneExtension.get()) or .extension(ElasticsearchExtension.get()) in the query DSL to retrieve a LuceneSearchQuery or an ElasticsearchQuery, which expose these explain() methods.

For more information, refer to the javadoc of LuceneSearchQuery and ElasticsearchQuery.

Experimental support for Bytecode-enhanced entities

In HSEARCH-3581 we introduced experimental support for extracting information from lazy properties of bytecode-enhanced entities.

We are looking for feedback: if you use bytecode enhancement, give it a try and please us know of any problem.

Other improvements and bug fixes

  • HSEARCH-3573: The .object() projection has been renamed to .entity().

  • HSEARCH-3577: Longs can no longer be passed to .fetch() and .fetchHits query methods, because neither Lucene nor Elasticsearch supports it. Use integers instead.

  • HSEARCH-3571: The syntax of requests sent to Elasticsearch has been slightly altered to avoid warnings with Elasticsearch 6.7 and above.

  • HSEARCH-3539: Lucene distance projections are now thread safe.

  • HSEARCH-3324: Multi-valued fields must now be declared. This should only affect you if you use a custom TypeBridge or PropertyBridge; otherwise, Hibernate Search will detect multi-valued fields automatically.

  • HSEARCH-3213: @Spatial bridge set at type level are no longer applied regardless of @IndexedEmbedded.includePaths.

  • HSEARCH-2663: Null handling for multi-valued properties is now consistent with single-valued properties.

  • HSEARCH-1857: An exception will now be thrown on usage of a SearchSession whose underlying ORM Session is closed.

  • HSEARCH-1645: The @ProvidedId annotation has now been officially removed.

And more. For a full list of changes since the previous releases, please see the release notes.

How to get this release

All details are available and up to date on the dedicated page on hibernate.org.

Feedback, issues, ideas?

To get in touch, use the following channels:


Back to top