Hi,

I understand that Hibernate Search searches the indexed files and proceed to retrieve managed objects from the db based on the ID.
- May I know what is the reason/benefit for such move? to avoid mapping of association since lucene cannot support associations.
- How does HS retrieve from the db based on the ID? by batch or individual ID? (would like to know how the db connections, thus, performance affected)



I also understand the use of Projection will enforce the search to only retrieve from indexed files. However, the results would be an array of objects instead of the managed objects. I have found this example that could convert the array of objects back to managed objects:

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() );
}

Pending to the above example,
1. Say, if Class A is a simple class containing no object attributes, Class A could be obtained as a managed object using Projection with the aid of ResultTransformer for all the attributes?
2. Say, Class A contain Class B, I use @IndexedEmbedded to annotate Class B. How should I set the projection so that I could obtained back Class A together with the embedded Class B?

The reason is that as result of the search you want to have Hibernate managed objects. Meaning you can edit or delete the returned objects within your sesssion and the changes get persisted back into the db on transaction commit.


That depends on the query and whether you query a single index or multiple. Generally Search will try to load in batch, for example with a Criteria query using Restrictions.in.


Not quite. The example gives you back objects, but they are not managed, meaning they exist outside the Session scope.


--Hardy

Thanks for your prompt reply.

Regarding my first question, do you mean if the object is built from indexed files, it is not attached to the session, thus, could not be easily change and persist back to the db?

Could not see your reply for the final portion..

Exactly


No. You can create Class instances using projection and result transformer, but they are not managed.


You are free to project any fields you store in the index. Just make sure to get all fields you need to re-hydrate the objects. I recommend to build a little sample index and then open the index using the Lucene tool Luke. Looking at how the actual index looks like will help you to understand projection.

--Hardy

As i'm using the objects for displaying purposes, thus, I don't see a big issue that it is not managed.

However, I don't know how to "map" back the Class B attributes using ResultTransformer.
let's say I have projected: ClassA.w, ClassA.x, ClassB.y, ClassB.z

query.setResultTransformer(
new StaticAliasToBeanResultTransformer( ClassA.class, "w", "x", "ClassB.y", "ClassB.z" )
);

??

For pure displaying purposes having unmanaged objects should be fine. If the name of the association for Class B in Class A is b you have to project b.y and b.z. I am not sure where you get the StaticAliasToBeanResultTransformer from and whether it can handle this. In the worst case you have to write your own result transformer.

--Hardy

Thanks so much! It has been a great help!