3.7. Hibernate Search

Numbers and dates are now indexed as numeric fields by default. Properties of type Date, Calendar, int, long, float, double, and their corresponding wrappers, are no longer indexed as strings; instead, they use Lucene's appropriate numeric encoding. If any query previously targeted a string encoded field, but is not encoded numerically, it will need to be updated. Numeric fields must be searched with a NumericRangeQuery; however, if the Search query DSL is in use it should generate the correct query. In addition, any fields targeted by faceting should now be string encoded.

When using an @IndexedEmbedded annotation to include fields from a related entity, previously the id of the related entity was included. These fields are no longer included by default, but may be enabled by using the includeEmbeddedObjectId attriubte of the @IndexedEmbedded annotation, as seen below:

@IndexedEmbedded(includeEmbeddedObjectId=true)

@IndexedEmbedded(includeEmbeddedObjectId=true)

The SortField API has been modified to use SortField.Type.${CLASSNAME} instead of the previous SortField.${CLASSNAME}. This behavior is highlighted below:

// Previous API call
fulltextQuery.setSort( new Sort( new SortField( "title", SortField.STRING ) ) );

// New API call
fulltextQuery.setSort( new Sort( new SortField( "title", SortField.Type.STRING ) ) );

// Previous API call
fulltextQuery.setSort( new Sort( new SortField( "title", SortField.STRING ) ) );

// New API call
fulltextQuery.setSort( new Sort( new SortField( "title", SortField.Type.STRING ) ) );

Similar to Sorting Query Results, each target field must match an appropriate kind of Faceting Query. For instance, a Numeric Facet Query must match a field that was indexed exclusively via the matching Numeric type.

When using @Field(indexNullAs=) to encode a null marker value in the index, the type of marker must be compatible with all other values which are indexed in the same field. Previously a string keyword, such as null, could be encoded with numeric fields; however, this is no longer allowed. A number must now be used to represent the null value, such as -1.

Many of the Analyzer, TokenFilter, and CharFilter classes have been moved, and it is recommended to examine Lucene's API documentation to find the correct replacement.