Red Hat Training

A Red Hat training course is available for Red Hat JBoss Enterprise Application Platform

14.3. Querying

Hibernate Search can execute Lucene queries and retrieve domain objects managed by an Hibernate session. The search provides the power of Lucene without leaving the Hibernate paradigm, giving another dimension to the Hibernate classic search mechanisms (HQL, Criteria query, native SQL query).
Preparing and executing a query consists of following four steps:
  • Creating a FullTextSession
  • Creating a Lucene query using either Hibernate Search query DSL (recommended) or using the Lucene Query API
  • Wrapping the Lucene query using an org.hibernate.Query
  • Executing the search by calling for example list() or scroll()
To access the querying facilities, use a FullTextSession. This Search specific session wraps a regular org.hibernate.Session in order to provide query and indexing capabilities.

Example 14.30. Creating a FullTextSession

Session session = sessionFactory.openSession();
...
FullTextSession fullTextSession = Search.getFullTextSession(session);
Use the FullTextSession to build a full-text query using either the Hibernate Search query DSL or the native Lucene query.
Use the following code when using the Hibernate Search query DSL:
final QueryBuilder b = fullTextSession.getSearchFactory().buildQueryBuilder().forEntity( Myth.class ).get();

org.apache.lucene.search.Query luceneQuery =
    b.keyword()
        .onField("history").boostedTo(3)
        .matching("storm")
        .createQuery();

org.hibernate.Query fullTextQuery = fullTextSession.createFullTextQuery( luceneQuery );
List result = fullTextQuery.list(); //return a list of managed objects
As an alternative, write the Lucene query using either the Lucene query parser or the Lucene programmatic API.

Example 14.31. Creating a Lucene query via the QueryParser

SearchFactory searchFactory = fullTextSession.getSearchFactory();
org.apache.lucene.queryParser.QueryParser parser = 
    new QueryParser("title", searchFactory.getAnalyzer(Myth.class) );
try {
    org.apache.lucene.search.Query luceneQuery = parser.parse( "history:storm^3" );
}
catch (ParseException e) {
    //handle parsing failure
}

org.hibernate.Query fullTextQuery = fullTextSession.createFullTextQuery(luceneQuery);
List result = fullTextQuery.list(); //return a list of managed objects
A Hibernate query built on the Lucene query is a org.hibernate.Query. This query remains in the same paradigm as other Hibernate query facilities, such as HQL (Hibernate Query Language), Native, and Criteria. Use methods such as list(), uniqueResult(), iterate() and scroll() with the query.
The same extensions are available with the Hibernate Java Persistence APIs:

Example 14.32. Creating a Search query using the JPA API

EntityManager em = entityManagerFactory.createEntityManager();

FullTextEntityManager fullTextEntityManager = 
    org.hibernate.search.jpa.Search.getFullTextEntityManager(em);

...
final QueryBuilder b = fullTextEntityManager.getSearchFactory()
    .buildQueryBuilder().forEntity( Myth.class ).get();

org.apache.lucene.search.Query luceneQuery =
    b.keyword()
        .onField("history").boostedTo(3)
        .matching("storm")
        .createQuery();
javax.persistence.Query fullTextQuery = fullTextEntityManager.createFullTextQuery( luceneQuery );

List result = fullTextQuery.getResultList(); //return a list of managed objects

Note

In these examples, the Hibernate API has been used. The same examples can also be written with the Java Persistence API by adjusting the way the FullTextQuery is retrieved.

14.3.1. Building Queries

Hibernate Search queries are built on Lucene queries, allowing users to use any Lucene query type. When the query is built, Hibernate Search uses org.hibernate.Query as the query manipulation API for further query processing.

14.3.1.1. Building a Lucene Query Using the Lucene API

With the Lucene API, use either the query parser (simple queries) or the Lucene programmatic API (complex queries). Building a Lucene query is out of scope for the Hibernate Search documentation. For details, see the online Lucene documentation or a copy of Lucene in Action or Hibernate Search in Action.

14.3.1.2. Building a Lucene Query

The Lucene programmatic API enables full-text queries. However, when using the Lucene programmatic API, the parameters must be converted to their string equivalent and must also apply the correct analyzer to the right field. A ngram analyzer for example uses several ngrams as the tokens for a given word and should be searched as such. It is recommended to use the QueryBuilder for this task.
The Hibernate Search query API is fluent, with the following key characteristics:
  • Method names are in English. As a result, API operations can be read and understood as a series of English phrases and instructions.
  • It uses IDE autocompletion which helps possible completions for the current input prefix and allows the user to choose the right option.
  • It often uses the chaining method pattern.
  • It is easy to use and read the API operations.
To use the API, first create a query builder that is attached to a given indexedentitytype. This QueryBuilder knows what analyzer to use and what field bridge to apply. Several QueryBuilders (one for each entity type involved in the root of your query) can be created. The QueryBuilder is derived from the SearchFactory.
QueryBuilder mythQB = searchFactory.buildQueryBuilder().forEntity( Myth.class ).get();
The analyzer used for a given field or fields can also be overridden.
QueryBuilder mythQB = searchFactory.buildQueryBuilder()
    .forEntity( Myth.class )
        .overridesForField("history","stem_analyzer_definition")
    .get();
The query builder is now used to build Lucene queries. Customized queries generated using Lucene's query parser or Query objects assembled using the Lucene programmatic API are used with the Hibernate Search DSL.

14.3.1.3. Keyword Queries

The following example shows how to search for a specific word:
Query luceneQuery = mythQB.keyword().onField("history").matching("storm").createQuery();

Table 14.4. Keyword query parameters

Parameter Description
keyword() Use this parameter to find a specific word
onField() Use this parameter to specify in which lucene field to search the word
matching() use this parameter to specify the match for search string
createQuery() creates the Lucene query object
  • The value "storm" is passed through the history FieldBridge. This is useful when numbers or dates are involved.
  • The field bridge value is then passed to the analyzer used to index the field history. This ensures that the query uses the same term transformation than the indexing (lower case, ngram, stemming and so on). If the analyzing process generates several terms for a given word, a boolean query is used with the SHOULD logic (roughly an OR logic).
To search a property that is not of type string.
@Indexed 
public class Myth {
  @Field(analyze = Analyze.NO) 
  @DateBridge(resolution = Resolution.YEAR)
  public Date getCreationDate() { return creationDate; }
  public Date setCreationDate(Date creationDate) { this.creationDate = creationDate; }
  private Date creationDate;
  
  ...
}

Date birthdate = ...;
Query luceneQuery = mythQb.keyword().onField("creationDate").matching(birthdate).createQuery();

Note

In plain Lucene, the Date object had to be converted to its string representation (in this case the year)
This conversion works for any object, provided that the FieldBridge has an objectToString method (and all built-in FieldBridge implementations do).
The next example searches a field that uses ngram analyzers. The ngram analyzers index succession of ngrams of words, which helps to avoid user typos. For example, the 3-grams of the word hibernate are hib, ibe, ber, ern, rna, nat, ate.
@AnalyzerDef(name = "ngram",
  tokenizer = @TokenizerDef(factory = StandardTokenizerFactory.class ),
  filters = {
    @TokenFilterDef(factory = StandardFilterFactory.class),
    @TokenFilterDef(factory = LowerCaseFilterFactory.class),
    @TokenFilterDef(factory = StopFilterFactory.class),
    @TokenFilterDef(factory = NGramFilterFactory.class,
      params = { 
        @Parameter(name = "minGramSize", value = "3"),
        @Parameter(name = "maxGramSize", value = "3") } )
  }
)

public class Myth {
  @Field(analyzer=@Analyzer(definition="ngram") 
  public String getName() { return name; }
  public String setName(String name) { this.name = name; }
  private String name;
  
  ...
}

Date birthdate = ...;
Query luceneQuery = mythQb.keyword().onField("name").matching("Sisiphus")
   .createQuery();
The matching word "Sisiphus" will be lower-cased and then split into 3-grams: sis, isi, sip, iph, phu, hus. Each of these ngram will be part of the query. The user is then able to find the Sysiphus myth (with a y). All that is transparently done for the user.

Note

If the user does not want a specific field to use the field bridge or the analyzer then the ignoreAnalyzer() or ignoreFieldBridge() functions can be called.
To search for multiple possible words in the same field, add them all in the matching clause.
//search document with storm or lightning in their history
Query luceneQuery = 
    mythQB.keyword().onField("history").matching("storm lightning").createQuery();
To search the same word on multiple fields, use the onFields method.
Query luceneQuery = mythQB
    .keyword()
    .onFields("history","description","name")
    .matching("storm")
    .createQuery();
Sometimes, one field should be treated differently from another field even if searching the same term, use the andField() method for that.
Query luceneQuery = mythQB.keyword()
    .onField("history")
    .andField("name")
      .boostedTo(5)
    .andField("description")
    .matching("storm")
    .createQuery();
In the previous example, only field name is boosted to 5.

14.3.1.4. Fuzzy Queries

To execute a fuzzy query (based on the Levenshtein distance algorithm), start with a keyword query and add the fuzzy flag.
Query luceneQuery = mythQB
    .keyword()
      .fuzzy()
        .withThreshold( .8f )
        .withPrefixLength( 1 )
    .onField("history")
    .matching("starm")
    .createQuery();
The threshold is the limit above which two terms are considering matching. It is a decimal between 0 and 1 and the default value is 0.5. The prefixLength is the length of the prefix ignored by the "fuzzyness". While the default value is 0, a nonzero value is recommended for indexes containing a huge number of distinct terms.

14.3.1.5. Wildcard Queries

Wildcard queries are useful in circumstances where only part of the word is known. The ? represents a single character and * represents multiple characters. Note that for performance purposes, it is recommended that the query does not start with either ? or *.
Query luceneQuery = mythQB
    .keyword()
      .wildcard()
    .onField("history")
    .matching("sto*")
    .createQuery();

Note

Wildcard queries do not apply the analyzer on the matching terms. The risk of * or ? being mangled is too high.

14.3.1.6. Phrase Queries

So far we have been looking for words or sets of words, the user can also search exact or approximate sentences. Use phrase() to do so.
Query luceneQuery = mythQB
    .phrase()
    .onField("history")
    .sentence("Thou shalt not kill")
    .createQuery();
Approximate sentences can be searched by adding a slop factor. The slop factor represents the number of other words permitted in the sentence: this works like a within or near operator.
Query luceneQuery = mythQB
    .phrase()
      .withSlop(3)
    .onField("history")
    .sentence("Thou kill")
    .createQuery();

14.3.1.7. Range Queries

A range query searches for a value in between given boundaries (included or not) or for a value below or above a given boundary (included or not).
//look for 0 <= starred < 3
Query luceneQuery = mythQB
    .range()
    .onField("starred")
    .from(0).to(3).excludeLimit()
    .createQuery();

//look for myths strictly BC
Date beforeChrist = ...;
Query luceneQuery = mythQB
    .range()
    .onField("creationDate")
    .below(beforeChrist).excludeLimit()
    .createQuery();

14.3.1.8. Combining Queries

Queries can be aggregated (combined) to create more complex queries. The following aggregation operators are available:
  • SHOULD: the query should contain the matching elements of the subquery.
  • MUST: the query must contain the matching elements of the subquery.
  • MUST NOT: the query must not contain the matching elements of the subquery.
The subqueries can be any Lucene query including a boolean query itself.

Example 14.33.  MUST NOT Query

//look for popular modern myths that are not urban
Date twentiethCentury = ...;
Query luceneQuery = mythQB
    .bool()
      .must( mythQB.keyword().onField("description").matching("urban").createQuery() )
        .not()
      .must( mythQB.range().onField("starred").above(4).createQuery() )
      .must( mythQB
        .range()
        .onField("creationDate")
        .above(twentiethCentury)
        .createQuery() )
    .createQuery();

Example 14.34.  SHOULD Query

//look for popular myths that are preferably urban
Query luceneQuery = mythQB
    .bool()
      .should( mythQB.keyword().onField("description").matching("urban").createQuery() )
      .must( mythQB.range().onField("starred").above(4).createQuery() )
    .createQuery();

Example 14.35.  NOT Query

//look for all myths except religious ones
Query luceneQuery = mythQB
    .all()
      .except( monthQb
        .keyword()
        .onField( "description_stem" )
        .matching( "religion" )
        .createQuery() 
      )
    .createQuery();

14.3.1.9. Query Options

The Hibernate Search query DSL is an easy to use and easy to read query API. In accepting and producing Lucene queries, you can incorporate query types not yet supported by the DSL.
The following is a summary of query options for query types and fields:
  • boostedTo (on query type and on field) boosts the whole query or the specific field to a given factor.
  • withConstantScore (on query) returns all results that match the query have a constant score equals to the boost.
  • filteredBy(Filter)(on query) filters query results using the Filter instance.
  • ignoreAnalyzer (on field) ignores the analyzer when processing this field.
  • ignoreFieldBridge (on field) ignores field bridge when processing this field.

Example 14.36. Combination of Query Options

Query luceneQuery = mythQB
    .bool()
      .should( mythQB.keyword().onField("description").matching("urban").createQuery() )
      .should( mythQB
        .keyword()
        .onField("name")
          .boostedTo(3)
          .ignoreAnalyzer()
        .matching("urban").createQuery() )
      .must( mythQB
        .range()
          .boostedTo(5).withConstantScore()
        .onField("starred").above(4).createQuery() )
    .createQuery();

14.3.1.10. Build a Hibernate Search Query

14.3.1.10.1. Generality
After building the Lucene query, wrap it within a Hibernate query. The query searches all indexed entities and returns all types of indexed classes unless explicitly configured not to do so.

Example 14.37. Wrapping a Lucene Query in a Hibernate Query

FullTextSession fullTextSession = Search.getFullTextSession( session );
org.hibernate.Query fullTextQuery = fullTextSession.createFullTextQuery( luceneQuery );
For improved performance, restrict the returned types as follows:

Example 14.38. Filtering the Search Result by Entity Type

fullTextQuery = fullTextSession
    .createFullTextQuery( luceneQuery, Customer.class );

// or

fullTextQuery = fullTextSession
    .createFullTextQuery( luceneQuery, Item.class, Actor.class );
The first part of the second example only returns the matching Customers. The second part of the same example returns matching Actors and Items. The type restriction is polymorphic. As a result, if the two subclasses Salesman and Customer of the base class Person return, specify Person.class to filter based on result types.
14.3.1.10.2. Pagination
To avoid performance degradation, it is recommended to restrict the number of returned objects per query. A user navigating from one page to another page is a very common use case. The way to define pagination is similar to defining pagination in a plain HQL or Criteria query.

Example 14.39. Defining pagination for a search query

org.hibernate.Query fullTextQuery = 
    fullTextSession.createFullTextQuery( luceneQuery, Customer.class );
fullTextQuery.setFirstResult(15); //start from the 15th element
fullTextQuery.setMaxResults(10); //return 10 elements

Note

It is still possible to get the total number of matching elements regardless of the pagination via fulltextQuery.getResultSize()
14.3.1.10.3. Sorting
Apache Lucene contains a flexible and powerful result sorting mechanism. The default sorting is by relevance and is appropriate for a large variety of use cases. The sorting mechanism can be changed to sort by other properties using the Lucene Sort object to apply a Lucene sorting strategy.

Example 14.40. Specifying a Lucene Sort

org.hibernate.search.FullTextQuery query = s.createFullTextQuery( query, Book.class );
org.apache.lucene.search.Sort sort = new Sort(
    new SortField("title", SortField.STRING));
query.setSort(sort);
List results = query.list();

Note

Fields used for sorting must not be tokenized. For more information about tokenizing, see Section 14.2.1.1.2, “@Field”.
14.3.1.10.4. Fetching Strategy
Hibernate Search loads objects using a single query if the return types are restricted to one class. Hibernate Search is restricted by the static fetching strategy defined in the domain model. It is useful to refine the fetching strategy for a specific use case as follows:

Example 14.41. Specifying FetchMode on a query

Criteria criteria = 
    s.createCriteria( Book.class ).setFetchMode( "authors", FetchMode.JOIN );
s.createFullTextQuery( luceneQuery ).setCriteriaQuery( criteria );
In this example, the query will return all Books matching the LuceneQuery. The authors collection will be loaded from the same query using an SQL outer join.
In a criteria query definition, the type is guessed based on the provided criteria query. As a result, it is not necessary to restrict the return entity types.

Important

The fetch mode is the only adjustable property. Do not use a restriction (a where clause) on the Criteria query because the getResultSize() throws a SearchException if used in conjunction with a Criteria with restriction.
If more than one entity is expected, do not use setCriteriaQuery.
14.3.1.10.5. Projection
In some cases, only a small subset of the properties is required. Use Hibernate Search to return a subset of properties as follows:
Hibernate Search extracts properties from the Lucene index and converts them to their object representation and returns a list of Object[]. Projections prevent a time consuming database round-trip. However, they have following constraints:
  • The properties projected must be stored in the index (@Field(store=Store.YES)), which increases the index size.
  • The properties projected must use a FieldBridge implementing org.hibernate.search.bridge.TwoWayFieldBridge or org.hibernate.search.bridge.TwoWayStringBridge, the latter being the simpler version.

    Note

    All Hibernate Search built-in types are two-way.
  • Only the simple properties of the indexed entity or its embedded associations can be projected. Therefore a whole embedded entity cannot be projected.
  • Projection does not work on collections or maps which are indexed via @IndexedEmbedded
Lucene provides metadata information about query results. Use projection constants to retrieve the metadata.

Example 14.42. Using Projection to Retrieve Metadata

org.hibernate.search.FullTextQuery query = 
    s.createFullTextQuery( luceneQuery, Book.class );
query.setProjection( FullTextQuery.SCORE, FullTextQuery.THIS, "mainAuthor.name" );
List results = query.list();
Object[] firstResult = (Object[]) results.get(0);
float score = firstResult[0];
Book book = firstResult[1];
String authorName = firstResult[2];
Fields can be mixed with the following projection constants:
  • FullTextQuery.THIS: returns the initialized and managed entity (as a non projected query would have done).
  • FullTextQuery.DOCUMENT: returns the Lucene Document related to the object projected.
  • FullTextQuery.OBJECT_CLASS: returns the class of the indexed entity.
  • FullTextQuery.SCORE: returns the document score in the query. Scores are handy to compare one result against an other for a given query but are useless when comparing the result of different queries.
  • FullTextQuery.ID: the ID property value of the projected object.
  • FullTextQuery.DOCUMENT_ID: the Lucene document ID. Be careful in using this value as a Lucene document ID can change over time between two different IndexReader opening.
  • FullTextQuery.EXPLANATION: returns the Lucene Explanation object for the matching object/document in the given query. This is not suitable for retrieving large amounts of data. Running explanation typically is as costly as running the whole Lucene query per matching element. As a result, projection is recommended.
14.3.1.10.6. Customizing Object Initialization Strategies
By default, Hibernate Search uses the most appropriate strategy to initialize entities matching the full text query. It executes one (or several) queries to retrieve the required entities. This approach minimizes database trips where few of the retrieved entities are present in the persistence context (the session) or the second level cache.
If entities are present in the second level cache, force Hibernate Search to look into the cache before retrieving a database object.

Example 14.43. Check the second-level cache before using a query

FullTextQuery query = session.createFullTextQuery(luceneQuery, User.class);
query.initializeObjectWith(
    ObjectLookupMethod.SECOND_LEVEL_CACHE,
    DatabaseRetrievalMethod.QUERY
);
ObjectLookupMethod defines the strategy to check if an object is easily accessible (without fetching it from the database). Other options are:
  • ObjectLookupMethod.PERSISTENCE_CONTEXT is used if many matching entities are already loaded into the persistence context (loaded in the Session or EntityManager).
  • ObjectLookupMethod.SECOND_LEVEL_CACHE checks the persistence context and then the second-level cache.
Set the following to search in the second-level cache:
  • Correctly configure and activate the second-level cache.
  • Enable the second-level cache for the relevant entity. This is done using annotations such as @Cacheable.
  • Enable second-level cache read access for either Session, EntityManager or Query. Use CacheMode.NORMAL in Hibernate native APIs or CacheRetrieveMode.USE in Java Persistence APIs.

Warning

Unless the second-level cache implementation is EHCache or Infinispan, do not use ObjectLookupMethod.SECOND_LEVEL_CACHE. Other second-level cache providers do not implement this operation efficiently.
Customize how objects are loaded from the database using DatabaseRetrievalMethod as follows:
  • QUERY (default) uses a set of queries to load several objects in each batch. This approach is recommended.
  • FIND_BY_ID loads one object at a time using the Session.get or EntityManager.find semantic. This is recommended if the batch size is set for the entity, which allows Hibernate Core to load entities in batches.
14.3.1.10.7. Limiting the Time of a Query
Limit the time a query takes in Hibernate Guide as follows:
  • Raise an exception when arriving at the limit.
  • Limit to the number of results retrieved when the time limit is raised.
14.3.1.10.8. Raise an Exception on Time Limit
If a query uses more than the defined amount of time, a QueryTimeoutException is raised (org.hibernate.QueryTimeoutException or javax.persistence.QueryTimeoutException depending on the programmatic API).
To define the limit when using the native Hibernate APIs, use one of the following approaches:

Example 14.44. Defining a Timeout in Query Execution

Query luceneQuery = ...;
FullTextQuery query = fullTextSession.createFullTextQuery(luceneQuery, User.class);

//define the timeout in seconds
query.setTimeout(5);

//alternatively, define the timeout in any given time unit
query.setTimeout(450, TimeUnit.MILLISECONDS);

try {
    query.list();
}
catch (org.hibernate.QueryTimeoutException e) {
    //do something, too slow
}
The getResultSize(), iterate() and scroll() honor the timeout until the end of the method call. As a result, Iterable or the ScrollableResults ignore the timeout. Additionally, explain() does not honor this timeout period. This method is used for debugging and to check the reasons for slow performance of a query.
The following is the standard way to limit execution time using the Java Persistence API (JPA):

Example 14.45. Defining a Timeout in Query Execution

Query luceneQuery = ...;
FullTextQuery query = fullTextEM.createFullTextQuery(luceneQuery, User.class);

//define the timeout in milliseconds
query.setHint( "javax.persistence.query.timeout", 450 );

try {
    query.getResultList();
}
catch (javax.persistence.QueryTimeoutException e) {
    //do something, too slow
}

Important

The example code does not guarantee that the query stops at the specified results amount.

14.3.2. Retrieving the Results

After building the Hibernate query, it is executed the same way as a HQL or Criteria query. The same paradigm and object semantic apply to a Lucene Query query and the common operations like: list(), uniqueResult(), iterate(), scroll() are available.

14.3.2.1. Performance Considerations

If you expect a reasonable number of results (for example using pagination) and expect to work on all of them, list() or uniqueResult() are recommended. list() work best if the entity batch-size is set up properly. Note that Hibernate Search has to process all Lucene Hits elements (within the pagination) when using list() , uniqueResult() and iterate().
If you wish to minimize Lucene document loading, scroll() is more appropriate. Don't forget to close the ScrollableResults object when you're done, since it keeps Lucene resources. If you expect to use scroll, but wish to load objects in batch, you can use query.setFetchSize(). When an object is accessed, and if not already loaded, Hibernate Search will load the next fetchSize objects in one pass.

Important

Pagination is preferred over scrolling.

14.3.2.2. Result Size

It is sometimes useful to know the total number of matching documents:
  • to provide a total search results feature, as provided by Google searches. For example, "1-10 of about 888,000,000 results"
  • to implement a fast pagination navigation
  • to implement a multi-step search engine that adds approximation if the restricted query returns zero or not enough results
Of course it would be too costly to retrieve all the matching documents. Hibernate Search allows you to retrieve the total number of matching documents regardless of the pagination parameters. Even more interesting, you can retrieve the number of matching elements without triggering a single object load.

Example 14.46. Determining the Result Size of a Query

org.hibernate.search.FullTextQuery query = 
    s.createFullTextQuery( luceneQuery, Book.class );
//return the number of matching books without loading a single one
assert 3245 == query.getResultSize(); 

org.hibernate.search.FullTextQuery query = 
    s.createFullTextQuery( luceneQuery, Book.class );
query.setMaxResult(10);
List results = query.list();
//return the total number of matching books regardless of pagination
assert 3245 == query.getResultSize();

Note

Like Google, the number of results is approximation if the index is not fully up-to-date with the database (asynchronous cluster for example).

14.3.2.3. ResultTransformer

Projection results are returned as Object arrays. If the data structure used for the object does not match the requirements of the application, apply a ResultTransformer. The ResultTransformer builds the required data structure after the query execution.

Example 14.47. Using ResultTransformer with Projections

org.hibernate.search.FullTextQuery query = 
    s.createFullTextQuery( luceneQuery, Book.class );
query.setProjection( "title", "mainAuthor.name" );

query.setResultTransformer( new StaticAliasToBeanResultTransformer( BookView.class, "title", "author" ) );
List<BookView> results = (List<BookView>) query.list();
for(BookView view : results) {
    log.info( "Book: " + view.getTitle() + ", " + view.getAuthor() );
}
Examples of ResultTransformer implementations can be found in the Hibernate Core codebase.

14.3.2.4. Understanding Results

If the results of a query are not what you expected, the Luke tool is useful in understanding the outcome. However, Hibernate Search also gives you access to the Lucene Explanation object for a given result (in a given query). This class is considered fairly advanced to Lucene users but can provide a good understanding of the scoring of an object. You have two ways to access the Explanation object for a given result:
  • Use the fullTextQuery.explain(int) method
  • Use projection
The first approach takes a document ID as a parameter and return the Explanation object. The document ID can be retrieved using projection and the FullTextQuery.DOCUMENT_ID constant.

Warning

The Document ID is unrelated to the entity ID. Be careful not to confuse these concepts.
In the second approach you project the Explanation object using the FullTextQuery.EXPLANATION constant.

Example 14.48. Retrieving the Lucene Explanation Object Using Projection

FullTextQuery ftQuery = s.createFullTextQuery( luceneQuery, Dvd.class )
        .setProjection( 
             FullTextQuery.DOCUMENT_ID, 
             FullTextQuery.EXPLANATION, 
             FullTextQuery.THIS );
@SuppressWarnings("unchecked") List<Object[]> results = ftQuery.list();
for (Object[] result : results) {
    Explanation e = (Explanation) result[1];
    display( e.toString() );
}
Use the Explanation object only when required as it is roughly as expensive as running the Lucene query again.

14.3.3. Filters

Apache Lucene has a powerful feature that allows you to filter query results according to a custom filtering process. This is a very powerful way to apply additional data restrictions, especially since filters can be cached and reused. Use cases include:
  • security
  • temporal data (example, view only last month's data)
  • population filter (example, search limited to a given category)
Hibernate Search pushes the concept further by introducing the notion of parameterizable named filters which are transparently cached. For people familiar with the notion of Hibernate Core filters, the API is very similar:

Example 14.49. Enabling Fulltext Filters for a Query

fullTextQuery = s.createFullTextQuery( query, Driver.class );
fullTextQuery.enableFullTextFilter("bestDriver");
fullTextQuery.enableFullTextFilter("security").setParameter( "login", "andre" );
fullTextQuery.list(); //returns only best drivers where andre has credentials
In this example we enabled two filters on top of the query. You can enable (or disable) as many filters as you like.
Declaring filters is done through the @FullTextFilterDef annotation. This annotation can be on any @Indexed entity regardless of the query the filter is later applied to. This implies that filter definitions are global and their names must be unique. A SearchException is thrown in case two different @FullTextFilterDef annotations with the same name are defined. Each named filter has to specify its actual filter implementation.

Example 14.50. Defining and Implementing a Filter


@FullTextFilterDefs( {
    @FullTextFilterDef(name = "bestDriver", impl = BestDriversFilter.class), 
    @FullTextFilterDef(name = "security", impl = SecurityFilterFactory.class) 
})
public class Driver { ... }
public class BestDriversFilter extends org.apache.lucene.search.Filter {

    public DocIdSet getDocIdSet(IndexReader reader) throws IOException {
        OpenBitSet bitSet = new OpenBitSet( reader.maxDoc() );
        TermDocs termDocs = reader.termDocs( new Term( "score", "5" ) );
        while ( termDocs.next() ) {
            bitSet.set( termDocs.doc() );
        }
        return bitSet;
    }
}
BestDriversFilter is an example of a simple Lucene filter which reduces the result set to drivers whose score is 5. In this example the specified filter implements the org.apache.lucene.search.Filter directly and contains a no-arg constructor.
If your Filter creation requires additional steps or if the filter you want to use does not have a no-arg constructor, you can use the factory pattern:

Example 14.51. Creating a filter using the factory pattern


@FullTextFilterDef(name = "bestDriver", impl = BestDriversFilterFactory.class)
public class Driver { ... }

public class BestDriversFilterFactory {

    @Factory
    public Filter getFilter() {
        //some additional steps to cache the filter results per IndexReader
        Filter bestDriversFilter = new BestDriversFilter();
        return new CachingWrapperFilter(bestDriversFilter);
    }
}
Hibernate Search will look for a @Factory annotated method and use it to build the filter instance. The factory must have a no-arg constructor.
Named filters come in handy where parameters have to be passed to the filter. For example a security filter might want to know which security level you want to apply:

Example 14.52. Passing parameters to a defined filter

fullTextQuery = s.createFullTextQuery( query, Driver.class );
fullTextQuery.enableFullTextFilter("security").setParameter( "level", 5 );
Each parameter name should have an associated setter on either the filter or filter factory of the targeted named filter definition.

Example 14.53. Using parameters in the actual filter implementation

public class SecurityFilterFactory {
    private Integer level;

    /**
     * injected parameter
     */
    public void setLevel(Integer level) {
        this.level = level;
    }

    @Key public FilterKey getKey() {
        StandardFilterKey key = new StandardFilterKey();
        key.addParameter( level );
        return key;
    }

    @Factory
    public Filter getFilter() {
        Query query = new TermQuery( new Term("level", level.toString() ) );
        return new CachingWrapperFilter( new QueryWrapperFilter(query) );
    }
}
Note the method annotated @Key returns a FilterKey object. The returned object has a special contract: the key object must implement equals() / hashCode() so that two keys are equal if and only if the given Filter types are the same and the set of parameters are the same. In other words, two filter keys are equal if and only if the filters from which the keys are generated can be interchanged. The key object is used as a key in the cache mechanism.
@Key methods are needed only if:
  • the filter caching system is enabled (enabled by default)
  • the filter has parameters
In most cases, using the StandardFilterKey implementation will be good enough. It delegates the equals() / hashCode() implementation to each of the parameters equals and hashcode methods.
As mentioned before the defined filters are per default cached and the cache uses a combination of hard and soft references to allow disposal of memory when needed. The hard reference cache keeps track of the most recently used filters and transforms the ones least used to SoftReferences when needed. Once the limit of the hard reference cache is reached additional filters are cached as SoftReferences. To adjust the size of the hard reference cache, use hibernate.search.filter.cache_strategy.size (defaults to 128). For advanced use of filter caching, implement your own FilterCachingStrategy. The classname is defined by hibernate.search.filter.cache_strategy.
This filter caching mechanism should not be confused with caching the actual filter results. In Lucene it is common practice to wrap filters using the IndexReader around a CachingWrapperFilter. The wrapper will cache the DocIdSet returned from the getDocIdSet(IndexReader reader) method to avoid expensive recomputation. It is important to mention that the computed DocIdSet is only cachable for the same IndexReader instance, because the reader effectively represents the state of the index at the moment it was opened. The document list cannot change within an opened IndexReader. A different/new IndexReader instance, however, works potentially on a different set of Documents (either from a different index or simply because the index has changed), hence the cached DocIdSet has to be recomputed.
Hibernate Search also helps with this aspect of caching. Per default the cache flag of @FullTextFilterDef is set to FilterCacheModeType.INSTANCE_AND_DOCIDSETRESULTS which will automatically cache the filter instance as well as wrap the specified filter around a Hibernate specific implementation of CachingWrapperFilter. In contrast to Lucene's version of this class SoftReferences are used together with a hard reference count (see discussion about filter cache). The hard reference count can be adjusted using hibernate.search.filter.cache_docidresults.size (defaults to 5). The wrapping behaviour can be controlled using the @FullTextFilterDef.cache parameter. There are three different values for this parameter:
Value Definition
FilterCacheModeType.NONE No filter instance and no result is cached by Hibernate Search. For every filter call, a new filter instance is created. This setting might be useful for rapidly changing data sets or heavily memory constrained environments.
FilterCacheModeType.INSTANCE_ONLY The filter instance is cached and reused across concurrent Filter.getDocIdSet() calls. DocIdSet results are not cached. This setting is useful when a filter uses its own specific caching mechanism or the filter results change dynamically due to application specific events making DocIdSet caching in both cases unnecessary.
FilterCacheModeType.INSTANCE_AND_DOCIDSETRESULTS Both the filter instance and the DocIdSet results are cached. This is the default value.
Last but not least - why should filters be cached? There are two areas where filter caching shines:
Filters should be cached in the following situations:
  • the system does not update the targeted entity index often (in other words, the IndexReader is reused a lot)
  • the Filter's DocIdSet is expensive to compute (compared to the time spent to execute the query)

14.3.3.1. Using Filters in a Sharded Environment

In a sharded environment it is possible to execute queries on a subset of the available shards. This can be done in two steps:

Procedure 14.1.  Query a Subset of Index Shards

  1. Create a sharding strategy that does select a subset of IndexManagers depending on a filter configuration.
  2. Activate the filter at query time.

Example 14.54.  Query a Subset of Index Shards

In this example the query is run against a specific customer shard if the customer filter is activated.
public class CustomerShardingStrategy implements IndexShardingStrategy {
	
	 // stored IndexManagers in a array indexed by customerID
	 private IndexManager[] indexManagers;
	 
	 public void initialize(Properties properties, IndexManager[] indexManagers) {
	   this.indexManagers = indexManagers;
	 }
	
	 public IndexManager[] getIndexManagersForAllShards() {
	   return indexManagers;
	 }
	
	 public IndexManager getIndexManagerForAddition(
	     Class<?> entity, Serializable id, String idInString, Document document) {
	   Integer customerID = Integer.parseInt(document.getFieldable("customerID").stringValue());
	   return indexManagers[customerID];
	 }
	
	 public IndexManager[] getIndexManagersForDeletion(
	     Class<?> entity, Serializable id, String idInString) {
	   return getIndexManagersForAllShards();
	 }
	
	  /**
	  * Optimization; don't search ALL shards and union the results; in this case, we 
	  * can be certain that all the data for a particular customer Filter is in a single
	  * shard; simply return that shard by customerID.
	  */
	 public IndexManager[] getIndexManagersForQuery(
	     FullTextFilterImplementor[] filters) {
	   FullTextFilter filter = getCustomerFilter(filters, "customer");
	   if (filter == null) {
	     return getIndexManagersForAllShards();
	   }
	   else {
	     return new IndexManager[] { indexManagers[Integer.parseInt(
	       filter.getParameter("customerID").toString())] };
	   }
	 }
	
	 private FullTextFilter getCustomerFilter(FullTextFilterImplementor[] filters, String name) {
	   for (FullTextFilterImplementor filter: filters) {
	     if (filter.getName().equals(name)) return filter;
	   }
	   return null;
	 }
	}
In this example, if the filter named customer is present, only the shard dedicated to this customer is queried, otherwise, all shards are returned. A given Sharding strategy can react to one or more filters and depends on their parameters.
The second step is to activate the filter at query time. While the filter can be a regular filter (as defined in Section 14.3.3, “Filters”) which also filters Lucene results after the query, you can make use of a special filter that will only be passed to the sharding strategy (and is otherwise ignored).
To use this feature, specify the ShardSensitiveOnlyFilter class when declaring your filter.
@Indexed
@FullTextFilterDef(name="customer", impl=ShardSensitiveOnlyFilter.class)
public class Customer {
   ...
}

FullTextQuery query = ftEm.createFullTextQuery(luceneQuery, Customer.class);
query.enableFulltextFilter("customer").setParameter("CustomerID", 5);
@SuppressWarnings("unchecked")
List<Customer> results = query.getResultList();
Note that by using the ShardSensitiveOnlyFilter, you do not have to implement any Lucene filter. Using filters and sharding strategy reacting to these filters is recommended to speed up queries in a sharded environment.

14.3.4. Faceting

Faceted search is a technique which allows the results of a query to be divided into multiple categories. This categorization includes the calculation of hit counts for each category and the ability to further restrict search results based on these facets (categories). Example 14.55, “Search for Hibernate Search on Amazon” shows a faceting example. The search results in fifteen hits which are displayed on the main part of the page. The navigation bar on the left, however, shows the category Computers & Internet with its subcategories Programming, Computer Science, Databases, Software, Web Development, Networking and Home Computing. For each of these subcategories the number of books is shown matching the main search criteria and belonging to the respective subcategory. This division of the category Computers & Internet is one concrete search facet. Another one is for example the average customer review.

Example 14.55. Search for Hibernate Search on Amazon

In Hibernate Search, the classes QueryBuilder and FullTextQuery are the entry point into the faceting API. The former creates faceting requests and the latter accesses the FacetManager. The FacetManager applies faceting requests on a query and selects facets that are added to an existing query to refine search results. The examples use the entity Cd as shown in Example 14.56, “Entity Cd”:
Search for Hibernate Search on Amazon

Figure 14.1. Search for Hibernate Search on Amazon

Example 14.56. Entity Cd

@Indexed
public class Cd {

    private int id;

    @Fields( {
        @Field,
        @Field(name = "name_un_analyzed", analyze = Analyze.NO)
    })
    private String name;

    @Field(analyze = Analyze.NO)
    @NumericField
    private int price;

    Field(analyze = Analyze.NO)
    @DateBridge(resolution = Resolution.YEAR)
    private Date releaseYear;

    @Field(analyze = Analyze.NO)
    private String label;


// setter/getter
...

14.3.4.1. Creating a Faceting Request

The first step towards a faceted search is to create the FacetingRequest. Currently two types of faceting requests are supported. The first type is called discrete faceting and the second type range faceting request. In the case of a discrete faceting request you specify on which index field you want to facet (categorize) and which faceting options to apply. An example for a discrete faceting request can be seen in Example 14.57, “Creating a discrete faceting request”:

Example 14.57. Creating a discrete faceting request

QueryBuilder builder = fullTextSession.getSearchFactory()
    .buildQueryBuilder()
        .forEntity( Cd.class )
            .get();
FacetingRequest labelFacetingRequest = builder.facet()
    .name( "labelFaceting" )
    .onField( "label")
    .discrete()
    .orderedBy( FacetSortOrder.COUNT_DESC )
    .includeZeroCounts( false )
    .maxFacetCount( 1 )
    .createFacetingRequest();
When executing this faceting request a Facet instance will be created for each discrete value for the indexed field label. The Facet instance will record the actual field value including how often this particular field value occurs within the original query results. orderedBy, includeZeroCounts and maxFacetCount are optional parameters which can be applied on any faceting request. orderedBy allows to specify in which order the created facets will be returned. The default is FacetSortOrder.COUNT_DESC, but you can also sort on the field value or the order in which ranges were specified. includeZeroCount determines whether facets with a count of 0 will be included in the result (per default they are) and maxFacetCount allows to limit the maximum amount of facets returned.

Note

At the moment there are several preconditions an indexed field has to meet in order to apply faceting on it. The indexed property must be of type String, Date or a subtype of Number and null values should be avoided. Furthermore the property has to be indexed with Analyze.NO and in case of a numeric property @NumericField needs to be specified.
The creation of a range faceting request is quite similar except that we have to specify ranges for the field values we are faceting on. A range faceting request can be seen in Example 14.58, “Creating a range faceting request” where three different price ranges are specified. below and above can only be specified once, but you can specify as many from - to ranges as you want. For each range boundary you can also specify via excludeLimit whether it is included into the range or not.

Example 14.58. Creating a range faceting request

QueryBuilder builder = fullTextSession.getSearchFactory()
    .buildQueryBuilder()
        .forEntity( Cd.class )
            .get();
FacetingRequest priceFacetingRequest = builder.facet()
    .name( "priceFaceting" )
    .onField( "price" )
    .range()
    .below( 1000 )
    .from( 1001 ).to( 1500 )
    .above( 1500 ).excludeLimit()
    .createFacetingRequest();

14.3.4.2. Applying a Faceting Request

A faceting request is applied to a query via the FacetManager class which can be retrieved via the FullTextQuery class.
You can enable as many faceting requests as you like and retrieve them afterwards via getFacets() specifying the faceting request name. There is also a disableFaceting() method which allows you to disable a faceting request by specifying its name.

Example 14.59. Applying a faceting request

// create a fulltext query
Query luceneQuery = builder.all().createQuery(); // match all query
FullTextQuery fullTextQuery = fullTextSession.createFullTextQuery( luceneQuery, Cd.class );

// retrieve facet manager and apply faceting request
FacetManager facetManager = fullTextQuery.getFacetManager();
facetManager.enableFaceting( priceFacetingRequest );

// get the list of Cds 
List<Cd> cds = fullTextQuery.list();
...

// retrieve the faceting results
List<Facet> facets = facetManager.getFacets( "priceFaceting" );
...

14.3.4.3. Restricting Query Results

Last but not least, you can apply any of the returned Facets as additional criteria on your original query in order to implement a "drill-down" functionality. For this purpose FacetSelection can be utilized. FacetSelections are available via the FacetManager and allow you to select a facet as query criteria (selectFacets), remove a facet restriction (deselectFacets), remove all facet restrictions (clearSelectedFacets) and retrieve all currently selected facets (getSelectedFacets). Example 14.60, “Restricting query results via the application of a FacetSelection shows an example.

Example 14.60. Restricting query results via the application of a FacetSelection

// create a fulltext query
Query luceneQuery = builder.all().createQuery(); // match all query
FullTextQuery fullTextQuery = fullTextSession.createFullTextQuery( luceneQuery, clazz );

// retrieve facet manager and apply faceting request
FacetManager facetManager = fullTextQuery.getFacetManager();
facetManager.enableFaceting( priceFacetingRequest );

// get the list of Cd 
List<Cd> cds = fullTextQuery.list();
assertTrue(cds.size() == 10);

// retrieve the faceting results
List<Facet> facets = facetManager.getFacets( "priceFaceting" );
assertTrue(facets.get(0).getCount() == 2)

// apply first facet as additional search criteria
facetManager.getFacetGroup( "priceFaceting" ).selectFacets( facets.get( 0 ) );

// re-execute the query
cds = fullTextQuery.list();
assertTrue(cds.size() == 2);

14.3.5. Optimizing the Query Process

Query performance depends on several criteria:
  • The Lucene query.
  • The number of objects loaded: use pagination (always) or index projection (if needed).
  • The way Hibernate Search interacts with the Lucene readers: defines the appropriate reader strategy.
  • Caching frequently extracted values from the index: see Section 14.3.5.1, “Caching Index Values: FieldCache”

14.3.5.1. Caching Index Values: FieldCache

The primary function of a Lucene index is to identify matches to your queries. After the query is performed the results must be analyzed to extract useful information. Hibernate Search would typically need to extract the Class type and the primary key.
Extracting the needed values from the index has a performance cost, which in some cases might be very low and not noticeable, but in some other cases might be a good candidate for caching.
The requirements depend on the kind of Projections being used (see Section 14.3.1.10.5, “Projection”), as in some cases the Class type is not needed as it can be inferred from the query context or other means.
Using the @CacheFromIndex annotation you can experiment with different kinds of caching of the main metadata fields required by Hibernate Search:
import static org.hibernate.search.annotations.FieldCacheType.CLASS;
import static org.hibernate.search.annotations.FieldCacheType.ID;

@Indexed
@CacheFromIndex( { CLASS, ID } )
public class Essay {
    ...
It is possible to cache Class types and IDs using this annotation:
  • CLASS: Hibernate Search will use a Lucene FieldCache to improve peformance of the Class type extraction from the index.
    This value is enabled by default, and is what Hibernate Search will apply if you don't specify the @CacheFromIndex annotation.
  • ID: Extracting the primary identifier will use a cache. This is likely providing the best performing queries, but will consume much more memory which in turn might reduce performance.

Note

Measure the performance and memory consumption impact after warmup (executing some queries). Performance may improve by enabling Field Caches but this is not always the case.
Using a FieldCache has two downsides to consider:
  • Memory usage: these caches can be quite memory hungry. Typically the CLASS cache has lower requirements than the ID cache.
  • Index warmup: when using field caches, the first query on a new index or segment will be slower than when you don't have caching enabled.
With some queries the classtype won't be needed at all, in that case even if you enabled the CLASS field cache, this might not be used; for example if you are targeting a single class, obviously all returned values will be of that type (this is evaluated at each Query execution).
For the ID FieldCache to be used, the ids of targeted entities must be using a TwoWayFieldBridge (as all builting bridges), and all types being loaded in a specific query must use the fieldname for the id, and have ids of the same type (this is evaluated at each Query execution).