In a previous post I mentioned a content management system I built for dynamically creating email content to notify users of events. The system is delivered via a RESTful web service that is called from disparate websites and programs within our software ecosystem. The basic requirement of the content builder is that those calling systems only need to provide a few key values. From that information it can determine what database object to gather data from and what content templates to use for building the email notifications.
The design of the system rests on the Abstract Factory design pattern. This allows the system to determine at run time what objects to create. In doing this the system uses reflection in a number of ways, one I discussed in a previous post mentioned above. Another I’m going to share in this post.
The system is passed an object which contains all the possible key fields needed to determine what data and content are needed. Depending on which key fields are populated, the system will create the needed data objects and apply the correct rules and templates for constructing the template.
Example Class:
namespace Extensions { public static class ObjectExtensions { public static T GetPropertyValue<T>(this object sourceObject, string key) { if (sourceObject.GetType().GetProperty(key) == null) return default(T); return (T)sourceObject.GetType().GetProperty(key).GetValue(sourceObject, null); } } }
Explanation:
The class is static because I like to write these sort of methods as extensions. I find them easier to use as a developer and really we just want to extend this method to any object. The method returns the generic type T because we don’t know what type the property we’re looking up is until run time.
The method first checks to see if the property(key) being requested exists. If it does not we return the default value of the generic type. In my system this works fine. Returning a null for a string or a zero for an integer will net the same affect because it’s as good as it not being populated. In other implementation you may want to throw an error here.
If the property exists then the value is returned to the caller it’s all done. You might ask, what if the property they pass as the “key” does not have a type of “T”? I contemplated that, one option could be to just return the default of the type T in that scenario. Instead, I decided not to handle that and allow the .Net framework to bubble the error up to the caller and allow it to make a decision on what to do. I’m not a big fan of validating methods are being called correctly in my logic layer or adding too much error handling. If not done correctly errors can be masked from the calling program and hidden. Below are a few unit tests that show how the method works.
Unit Tests:
[TestClass] public class ObjectExtensionsTests { public class SourceObject { public int Id { get; set; } public string Ids { get; set; } } [TestMethod] public void ValueExistsAndIsReturned() { var s = new SourceObject {Id = 123456}; Assert.AreEqual(123456, s.GetPropertyValue<int>( "Id")); } [TestMethod] public void ValueDoesNotExistsAndDefaultIntValueIsReturned() { var s = new SourceObject { Id = 123456 }; Assert.AreEqual(0, s.GetPropertyValue<int>("Id2")); } [TestMethod] public void ValueDoesNotExistsAndDefaultStringValueIsReturned() { var s = new SourceObject { Ids = "123456" }; Assert.AreEqual(null, s.GetPropertyValue<string>("Id2")); } [TestMethod] public void WrongTypeErrorReturned() { var error = false; try { var s = new SourceObject { Id = 123456 }; Assert.AreEqual(null, s.GetPropertyValue<string>("Id")); } catch (Exception) { error = true; } Assert.AreEqual(true, error); } }
There you have it. A simple extension method to get the value of an unknown property on an unknown object.
Image may be NSFW.
Clik here to view.
Image may be NSFW.Clik here to view.
