Hi,

1. I'm a complete HSearch/Lucene beginner.
2. I have read both Hibernate in Action (2sd Ed) and Hibernate Search in action, searched the forums, but found (almost) nothing...
3. Versions of Hibernate are the latest (since it's a new project)

I'm in the process of creating a data model with some fields in several languages. For example :



"descriptions" is a map with the key being the locale and the value the description in that locale, eg {{"en","my product"}, {"fr","mon produit"}}.
So far so good, but now the problem is indexing each description with the right analyzer, ie for the language used for the current row.

1. I found on this forum a thread mentioning "AnalyzerDiscriminator", but it seems it cannot be applied to a Map, does it?
2. I'm thinking to convert Map<Locale,String> to List<I15nData> (I15nData being a simple structure with a locale and a description fields) and applying "AnalyzerDiscriminator" to I15nData.

Please can someone give me some input on this?
Thanks

David

You can apply an AnalyzerDiscriminator on every kind of field, but in this case you'll get only one Analyzer for the indexing of the whole Map; of course that's not what you need.


that's not going to work, the Analyzer being used at index time is selected by the field name, using an @IndexedEmbedded collection you're basically appending all information in the same field, but you need to keep the information for each language in separate fields if you want them to use different Analyzers..

You will need a @ClassBridge, so you can define new fields dynamically from the Locale,String value of each entity.

Did you consider how you are going to query this? If it's allowed to search on more than one language in one query you might want to avoid having multiple Analyzers: the user query has to be analyzed with one Analyzer, consistent with the one used for indexing the data you want to match.

Hi,

I guess your approach could work depending of course how you design I15nData.
Another approach could be to use a FieldBridge for the description map together with a Lucene PerFieldAnalyzerWrapper. If you for example want to index the description for each language into a separate field, eg description_en, description_de, etc, then you can define the appropriate analyzers for these fields names in the wrapper. Then in your custom FieldBridge indexing the map you just have make sure to create the right field names depending on the locale.

--Hardy

right I jumped to the @ClassBridge, but the @FieldBridge Hardy is describing should be a better approach.

Still let me know how you are going to write the queries, you will have to consider the different named fields at Query definition.

Thanks for our answers !
That's a lot to process for a newbie. I guess I didn't start with the easiest part...

@s.grinovero :
1. each query will be for a single language
2. i don't understand why I have to use @IndexedEmbedded on the map ?

@hardy.ferentschik :
1. well, I was refering to your post :) the only one talkingt about AnalyzerDiscriminator in the forum
https://forum.hibernate.org/viewtopic.php?f=9&t=996099&p=2410221&hilit=AnalyzerDiscriminator#p2410221
2. I'll (re)read the related parts of the doc and the book, and certainly come back to you :)

you don't need IndexedEmbedded; it was needed to map the collection as you were trying to do in the first post.
IndexedEmbedded is used to include in the index of this entity the fields from the related objects,
in this case you might annotate fields of the I15nData class so to get those fields included in the index of your root
entity.
This will however generate multiple copies of the same-named field, so you don't need the IndexedEmbedded
but a FieldBridge to be able to customize the field name for each I15nData instance.