RavenDB MultiMap Index - ravendb

Question

I am new to RavenDB. I am trying to use a multi map index feature, though I am not sure if it is the best approach to my problem. So I have three documents: Unit, Car, People.
Car document looks like this:
{
Id: "cars/123",
PersonId: "people/1235",
UnitId: "units/4321",
Make: "Toyota",
Model: "Prius"
}
People document looks like this:
{
Id: "people/1235",
FirstName: "test",
LastName: "test"
}
And unit doc:
{
Id: "units/4321",
Address: "blah blah"
}
This is an abbreviated example, in my real app there are way more fields, so data de-normalization would be my last resort.
I need to create and index that will have all of this three docuemnts joined in one document. Something like this:
{
CarId: "cars/123",
PersonId: "people/1235",
UnitId: "units/4321",
Make: "Toyota",
Model: "Prius"
FirstName: "test",
LastName: "test"
Address: "blah blah"
}
// same unit different person owns a different car
{
CarId: "cars/122",
PersonId: "people/1236",
UnitId: "units/4321",
Make: "Toyota",
Model: "4runner"
FirstName: "test",
LastName: "test"
Address: "blah blah"
}
In a relational database I would just use two joins to People and Unit tables by ids
and my car table would be an aggregate entity.
Here is the index definition that I have:
public class MyMultiIndex : AbstractMultiMapIndexCreationTask<JoinedDocument>
{
public MyMultiIndex()
{
// creating maps
AddMap<Car>(cars => cars.Select(e => new { e.CarId, e.Make, e.Model, PersonId = e.PersonId, UnitId = e.UnitId, FirstName = (null)string, LastName = (null)string, Address = (nul)string }));
AddMap<People>(people => people.Select(e => new { CarId = (string)null, Make = (string)null, Model = (string)null, PersonId = e.Id, UnitId = (null)string, FirstName = e.FirstName, LastName = e.LastName, Address = (nul)string }));
AddMap<Unit>(people => people.Select(e => new { CarId = (string)null, Make = (string)null, Model = (string)null, PersonId = (null)string, UnitId = e.null, FirstName = (nul)string , LastName = (nul)string , Address = e.Address }));
Reduce = results => from result in results
group result by result.CarId
into g
select new JoinedDocument
{
CarId = g.Key,
PersonId = g.First(e => e.CarId == g.Key).PersonId,
UnitId = g.First(e => e.CarId == g.Key).UnitId,
Model = g.First(e => e.CarId == g.Key).Model,
Make = g.First(e => e.CarId == g.Key).Make,
**// this never works. It is like result set does not contain anything with this personId. It looks like AddMap for people document did not work.**
FirstName = results.First(e => e.PersonId == g.First(ie => ie.CarId == g.Key).PersonId).FirstName,
**// this never works. It is like result set does not contain anything with this personId. It looks like AddMap for people document did not work.**
LastName = results.First(e => e.PersonId == g.First(ie => ie.CarId == g.Key).PersonId).LastName,
**// this never works. It is like result set does not contain anything with this personId. It looks like AddMap for unit document did not work.**
UnitAddress = results.First(e => e.UnitId == g.First(ie => ie.CarId == g.Key).UnitId).LastName,
};
Index(map => map.Model, FieldIndexing.Analyzed);
Index(map => map.Make, FieldIndexing.Analyzed);
Index(map => map.LastName, FieldIndexing.Analyzed);
Index(map => map.FirstName, FieldIndexing.Analyzed);
Index(map => map.Make, FieldIndexing.Analyzed);
Index(map => map.UnitAddress, FieldIndexing.Analyzed);
}
}
When RavenDb runs this index I see errors when it is trying to run the Reduce function I have provided. It throws error when I am trying to match a record where person's first name and last name exist, same happens with the unit.

Answer 1

It looks like you're trying to fit a document database with an object model that has relationships. This blog may help you:
Keeping a Domain Model Pure with RavenDB
Keep in mind that this isn't the recommended use of RavenDB, but sometimes it's necessary, and this is a good way to handle it.

Answer 2

Can you have a car with no owner? or an address with no resident? If its false in both cases, I would model the car and unit to be embedded inside a person. So the person becomes your aggregate root and to get to a car or a unit you must go through a person.