Thursday, January 24, 2013

Repository Pattern with Cross Table Objects


In a recent post, I mentioned one of the advantages of using Entity Framework in your application is that you can build a generic repository class very easily. Of course, its not realistic to think that all access to the data will be a single table at a time. Most often it’s the case where you need to return data that spans multiple tables.

I’m  going to show you how I created a simple repository class that spans tables.

Creating the Summary/DTO Object

The first thing I like to start with is to create the simple POCO object that will be used to transport the data. This is essential to define first so that you do not get caught up in data structures, but instead define the data as the application is going to need it.

In the case with my database, I have a table called “Avail” that contains a ton of foreign keys to a contact table. I needed to display this data, but instead of a bunch of foreign keys, I needed to display the actual names of people. etc.

I ended up defining the object as follows:


Creating the Repository Class

After the DTO Object is created, its just a matter of creating a repository class that manages this object. In my implementation of the repository, I have an IRepository Interface that all of the repositories use to promote ease of understanding when working with them:


On top of this, I created a separate IFullAvailRepository so that I could customize this down the road if necessary with custom calls. For now, its empty.


I then created the repository class an implement the interface:


Implementing the interface

Now the hard (easy?) part. Using the power of Entity Framework and LINQ to Entities, implementing the interface for a custom object is actually quite easy. The most important part is now the objects are retrieved from the data context. By using an IQueryable object, we can abstract out filling the custom object into a simple property, and the rest of the methods just leverage that property.


Using the repository

Using this implementation, the DB isn’t actually hit until needed, so we can do queries to retrieve objects or sets of objects and only the items requested are in memory, instead of loading all of the items and doing the queries there. For example, this unit test retrieves 1 specific item out of the database, and its super fast.




I’m new to the repository pattern, but so far I really like the abstraction is provides to the data layer. I’m sure I’ll run into issues in the future, but so far, its working great!

No comments:

Post a Comment