Correct to use ignoreFieldBridge for "subentity.field" DSL?

neek · **Joined:** Wed Sep 30, 2009 5:29 am **Posts:** 29

Hibernate 3.6.0, HSearch 3.3.0. I wonder if anyone could clarify if I am right to solve my problem by using ignoreFieldBridge(), when I actually would benefit from the bridging of the fields in question (using my PadNumberBridge).

I have a CategoryEntity which has indexed 'left' and 'right' properties. Other entities hold attributes of type CategoryEntity and index their fields using @IndexedEmbedded, so I have:

Code:

@Entity
class CategoryEntity {
   @Column(name="lft")
   @Field(name="left", index=Index.UN_TOKENIZED, store=Store.YES)
   @FieldBridge(impl=PadNumberBridge.class,
         params = { @Parameter(name="pad", value="4") }
   )
   private int left;
   ... same for 'right' ...
}

@Entity
class ItemEntity {
   @IndexedEmbedded(depth=1)
   private CategoryEntity category;
}

This results in each Lucene document for ItemEntity being represented with a "category.left" and "category.right" field. Luke shows an ItemEntity being indexed with, for example,

Quote:

category.left = 0004
category.right = 0012

Until now I've used Hibernate Search query building, with a @FilterDef on the CategoryEntity class allowing me to enable the filter on the FullTextSession and specify the left and right values easily as 'int' parameters. The conversion from 4 to "0004" was done for me.

Now, I'm trying to change my code to use Hibernate Search query DSL and a QueryBuilder to build the same basic query. I find that I have trouble specifying a search for ItemEntity's with specific category.left and category.right values. If I use some naive code like:

Code:

qb.range().onField("category.left").above(searchCat.getLeft()).createQuery())

I get an exception saying it doesn't know what FieldBridge to use for "category.left". I cannot see how to specify a search on a sub-entity's field.

I ran into ignoreFieldBridge(), and tried it thus:

Code:

qb.range().onField("category.left").ignoreFieldBridge().above(searchCat.getLeft()).createQuery())

This does generate a Lucene query, but does not pad the 'left' integer, so I get e.g. "category.left:[4 TO *]" and the search returns no results.

If I pad the number manually and pass the resulting string "0004" to the above() clause of the QueryBuilder, to make "category.left:[0004 TO *]", then the search executes correctly and returns results, however it feels wrong to do this.

So, is there a better way to specify a search on "category.left" of my ItemEntity's? I feel the documentation (5.1.2. Building a Lucene query with the Hibernate Search query DSL) is lacking guidance in this area.

Thanks!
Nick

sanne.grinovero · **Posted:** Sun Feb 06, 2011 8:03 am

Hi Nick,
thanks for spotting this, I think it would be nice to have the DSL to provide exactly what you was expecting, but I don't think the fieldBridge on indexedEmbedded is being taken in consideration in the current implementation.

Please open a JIRA about it; you're also welcome to provide a testcase and a prototype of a solution; It shouldn't be hard, we have the list of bridges per attribute of each entity, just the linked entities are not being searched for in this case.

neek · **Joined:** Wed Sep 30, 2009 5:29 am **Posts:** 29

Thanks. I've opened http://opensource.atlassian.com/project ... SEARCH-682. I'm extremely busy with work and a newborn baby.

I failed to create a simple test case for a recent JIRA (http://opensource.atlassian.com/project ... SEARCH-678) , I doubt very much I will be able to create one and propose a solution for this issue. I would have to study how to do it and I don't think I have the required time or energy. If you point at a template unit test .java that should work, I'll probably be able to knock up a unit test?

sanne.grinovero · **Posted:** Wed Feb 09, 2011 9:24 am

Hi, thanks for the JIRA and congratulations for the baby.

In case you have some time, this should facilitate things: ContributingtoHibernateSearch, ignore the last part about releasing it, jump straight to cloning the GIT repository and coding conventions.

Writing a testcase is very simple as long as you extend our SearchTestCase, this setups the basic things you need and has some methods you can override to customize configuration - especially which annotated entities are relevant for you test. Add one (or more) entities to use it;

org.hibernate.search.test.RamDirectoryTest is a good example to take as a base to write a new test.

sometimes you can find an existing test which is very close to what you need, you're free to add a new method to an existing test and reuse some model for your purpose.