Indexes: Indexing Polymorphic Data

By default, RavenDB indexes operate only on a specific entity type, or a Collection, that ignores the inheritance hierarchy.

For example, let's assume that we have the following inheritance hierarchy:

Figure 1: Polymorphic indexes

If we saved a Cat, it would have a collection set to "Cats" and if we saved a Dog, it would be in collection "Dogs".

If we wanted to index cats by name, we would write:

from cat in docs.Cats
select new { }

And for dogs:

from dog in docs.Dogs
select new { }

Although it works, each index would only give us results for the animal it has been defined on. But what if we wanted to query across all animals?

Multi-Map Indexes

The easiest way to do this is by writing a multi-map index like this one:

IndexDefinition indexDefinition = new IndexDefinition();
HashSet<String> maps = new HashSet<>();
maps.add("docs.Cats.Select(c => new { name =})");
maps.add("docs.Dogs.Select(c => new { name =})");
public static class Animals_ByName extends AbstractJavaScriptIndexCreationTask {
    public Animals_ByName() {
            "map('cats', function (c){ return {name:}})",
            "map('dogs', function (d){ return {name:}})"

And query it like this:

List<Animal> results = session
    .query(Animal.class, Query.index("Animals/ByName"))
    .whereEquals("name", "Mitzy")
from index 'Animals/ByName'
where name = 'Mitzy'

Other Options

Another option would be to modify the way we generate the Collection for subclasses of Animal, like this:

try (IDocumentStore store = new DocumentStore()) {
    store.getConventions().setFindCollectionName(clazz -> {
        if (Animal.class.isAssignableFrom(clazz)) {
            return "Animals";

        return DocumentConventions.defaultGetCollectionName(clazz);

Using this method, we can now index on all animals using:

from animal in docs.Animals
select new { }

But what happens when you don't want to modify the entity name of an entity itself?

You can create a polymorphic index using:

from animal in docs.WhereEntityIs("Cats", "Dogs")
select new { animal.Name }

It will generate an index that matches both Cats and Dogs.