SOLR中的自定义字段类

Question

SOLR allows use to define a field type using settings such as:

<fieldtype name= "text_th" class = "solr.TextField" />

Here the class is an in-built SOLR class. Are we allowed to define our own class and use it in a fieldType definition? I would like to have a class such as MyCompany.MyClass{ String name, int age, float salary }

This is so that the value that I store in the SOLR document is a composite value made up of 3 components. If this is allowed, are there any guidelines or defined practices for such classes such as how to develop a tokenizer for such a custom class, etc.?

Thanks,

Yash

Answer 1

Yes, you can create your own FieldType - there should however be very specific reasons to do so. The Solr backend is Lucene, so it'll be your own responsibility to marshal content between the Solr representation and the backing Lucene implementation.

Possible starting points could be the implementation of StrField in Solr (which is just a simple field in Lucene as well), or far more advanced examples such as LatLonType and PointField .

Be aware that the reason to create a new field type should be to express something that isn't possible with the currently available field types (or greatly simplify handling of such values). Remember that you might have to concern yourself with how sorting should work, how filtering should work, etc., and syntaxes for all these items.

Usually you're far better off (and an actually maintainable solution) by creating a separate collection, or by indexing your content in more than one way. Custom field types (on the Lucene / Solr level) is almost never the answer.