Proper way to handle nullable fields in RavenDB Map/Reduces? - dictionary

Question

How should I Map an object with a nullable field? I guess I must turn the nullable field into a non-nullable version, and it's that step that I stumble upon.
What is the proper way to map nullable properties?
public class Visit {
public string Id { get; set; }
public int? MediaSourceId { get; set; }
}
public class MapReduceResult
{
public string VisitId { get; set; }
public int MediaSourceId { get; set; }
public string Version { get; set; }
public int Count { get; set; }
}
AddMap<Visit>(
visits =>
from visit in visits
select new
{
VisitId = visit.Id,
MediaSourceId =
(visit.MediaSourceId.HasValue)
? visit.MediaSourceId
: UNUSED_MEDIASOURCE_ID,
Version = (string) null,
Count = 1
});
This doesn't work! In fact; this Map is completely ignored, while the other Maps work fine, and they are in the end Reduced as expected.
Thanks for helping me!
Below is a newly added test case that fails with a "Cannot assign <null> to anonymous type property". How am I supposed to get this flying with the least amount of pain?
[TestFixture]
public class MyIndexTest
{
private IDocumentStore _documentStore;
[SetUp]
public void SetUp()
{
_documentStore = new EmbeddableDocumentStore {RunInMemory = true}.Initialize();
_documentStore.DatabaseCommands.DisableAllCaching();
IndexCreation.CreateIndexes(typeof (MyIndex).Assembly, _documentStore);
}
[TearDown]
public void TearDown()
{
_documentStore.Dispose();
}
[Test]
public void ShouldWork()
{
InitData();
IList<MyIndex.MapReduceResult> mapReduceResults = null;
using (var session = _documentStore.OpenSession())
{
mapReduceResults =
session.Query<MyIndex.MapReduceResult>(
MyIndex.INDEX_NAME)
.Customize(x => x.WaitForNonStaleResults()).ToArray();
}
Assert.That(mapReduceResults.Count, Is.EqualTo(1));
}
private void InitData()
{
var visitOne = new Visit
{
Id = "visits/64",
MetaData = new MetaData {CreatedDate = new DateTime(1975, 8, 6, 0, 14, 0)},
MediaSourceId = 1,
};
var visitPageVersionOne = new VisitPageVersion
{
Id = "VisitPageVersions/123",
MetaData = new MetaData {CreatedDate = new DateTime(1975, 8, 6, 0, 14, 0)},
VisitId = "visits/64",
Version = "1"
};
using (var session = _documentStore.OpenSession())
{
session.Store(visitOne);
session.Store(visitPageVersionOne);
session.SaveChanges();
}
}
public class MyIndex :
AbstractMultiMapIndexCreationTask
<MyIndex.MapReduceResult>
{
public const string INDEX_NAME = "MyIndex";
public override string IndexName
{
get { return INDEX_NAME; }
}
public class MapReduceResult
{
public string VisitId { get; set; }
public int? MediaSourceId { get; set; }
public string Version { get; set; }
public int Count { get; set; }
}
public MyIndex()
{
AddMap<Visit>(
visits =>
from visit in visits
select new
{
VisitId = visit.Id,
MediaSourceId = (int?) visit.MediaSourceId,
Version = (string) null,
Count = 1
});
AddMap<VisitPageVersion>(
visitPageVersions =>
from visitPageVersion in visitPageVersions
select new
{
VisitId = visitPageVersion.VisitId,
MediaSourceId = (int?) null,
Version = visitPageVersion.Version,
Count = 0
});
Reduce =
results =>
from result in results
group result by result.VisitId
into g
select
new
{
VisitId = g.Key,
MediaSourceId = (int?) g.Select(x => x.MediaSourceId).FirstOrDefault(),
Version = g.Select(x => x.Version).FirstOrDefault(),
Count = g.Sum(x => x.Count)
};
}
}
}

Answer 1

You don't need to do anything to give nullables special treatment.
RavenDB will already take care of that.