Home > Agile, Extreme Programming, TDD, Test Driven Development > Refactoring Code: A Programmers Challenge

Refactoring Code: A Programmers Challenge

Kevin Jones: Refactoring

April 2nd 2009 | Kevin Jones

Refactoring Code: A Programmer’s Challenge

Today I encountered a question on the ASP.NET Forum that I got quite excited about, “How do I refactor this?” Refactoring is a big part of our daily coding practices and I was curious as to what other programmers would come up with. Here is the basic question re-worded a little bit for brevity:
I have a master page with a control on it. However, on certain pages I don’t want a control to be shown. At first it was just one or two pages, and my approach worked just fine. However, the number of pages I need to hide this control on has grown. Can anyone offer some advice on how to refactor this code?
This was his code:
string pagename = Request.ServerVariables[“PATH_INFO”].ToString().Trim().ToLower();
if (
pagename == “/index.aspx” ||
pagename == “/sitemap.aspx” ||
pagename == “/404.aspx” ||
pagename == “/status.aspx” ||
pagename == “/online_help.aspx”
)
{
SomePanel.Visible = false;
}

The original poster realized that his code wasn’t maintainable-a classic reason to refactor: Keep your code clean, readable, and maintainable.

The interesting thing about refactoring is that there is no de-facto way to improve this code-or any refactoring for that matter. Refactoring has a lot of patterns that have been used over the years, but how that pattern is implemented, and evolves, is a discussion that could go on forever. The key is: keep the code simple and keep the changes small. If we really wanted to, we could write over-the-top code; put these pages in a database table; and cache the table for performance. But if you refactor often enough the changes should always be incremental.

Here’s something else to consider: the ‘if’ statement will keep getting larger and larger every time a page is added. If I was implementing this I might have started in the same place. I wouldn’t have needed this control visible for this page, but then I might not have needed it for the next one either, or the next one. The original code is not bad-it just doesn’t meet the new criteria.
My suggestion is to remove the OR chain and store the page names in a local so one can just keep adding to a string array, and then check if it exists in the string array. This simple refactoring adds a little clarity to the code, but it’s not exactly a work of art:
string[] pagesToHideElement = new[]
{
“/index.aspx”,
“/sitemap.aspx”,
“/404.aspx”,
“/status.aspx”,
“/online_help.aspx”,
};
string pagename = Request.ServerVariables[“PATH_INFO”].ToString().Trim().ToLower();
if (Array.Exists(pagesToHideElement, s => s == pagename))
{
SomePanel.Visible = false;
}
We have isolated one of our problems: the growing list of pages we want to exclude. The developer can then keep adding to the array. This also provides flexibility we didn’t have before. We can now move our page list anywhere we like, like a new class and method. This is a very simple refactoring, but what else can we do? While the original person didn’t think of this, this code isn’t easy to test without a full-blown integration test.
Not to derail our refactoring discussion, but let’s talk briefly about testing and how it applies to refactoring. They go hand-in-hand, especially to an Agile developer. Refactoring is a necessary part of improving code, but how do you know that everything will still work? Many who oppose refactoring don’t write tests simply because they have no proof that a refactoring didn’t break the code. And refactoring isn’t always easy, especially when dealing with legacy code. Most Agile developers are familiar with the “Red-Green-Refactor” motto:

Refactoring Code

The diagram varies, but nevertheless it always says the same thing
1. Write a failing test first (That’s a core Test Driven Development value)
2. Make the test pass by implementing the required code
3. See the test pass
4. Refactor
5. Make sure you are still green
6. Improve test so that it is red again
It’s the fifth component that makes testing and refactoring go together well. Peace of mind comes with knowing that you didn’t break anything, assuming you have good testing practices-but of course you do!

Back to refactoring. How is this code refactoring tested without actually seeing if it shows the correct content or not? MVP or the MVC pattern are a good idea, but let’s start small. A pattern that I am fond of for these simple things is a decider class. The benefits of this pattern will become clear shortly.
This sort of decision-making logic doesn’t really belong in the presentation layer (UI layer). It can be deferred to a decider. The decider is exactly what it sounds like-a class that decides if the panel should be visible or not. So our decider can have a method on it that takes a page name, and return a Boolean value if the panel should be displayed. Simple enough, right? At this point we are extracting our logic out of the page and into a new class.
public class DisplayPanelDecider
{
public bool ShouldDisplayPanelOnPage(string pageName)
{
string[] pagesToHideElement = new[]
{
“/index.aspx”,
“/sitemap.aspx”,
“/404.aspx”,
“/status.aspx”,
“/online_help.aspx”,
};
return !Array.Exists(pagesToHideElement, s => s == pageName);
}
}
So now we have a piece of logic that we can test very easily. All that the ASPX page would have to do is create an instance and hand in the page name, but the core decision logic can be tested without any need for a browser.
This gives us a new power through. Now that we have this decision making logic in its own method, we can make our method do whatever we like. If, in a year’s time, this no longer feasible, we can alter our method to pull the page names from an XML file or a database. We’ll look at that later.


A quick recap of what we did:

  • We changed the functionality to give us a bit more functionality and moved from a large OR operation to a more dynamic approach.
  • We extracted our logic out to a new class.
  • Two simple changes…and the benefits are huge!
  • We can now test this logic very easily through our favorite testing framework. XUnit MSTest, etc. (I refer to XUnit as “Any of the testing frameworks, like NUnit, MbUnit, etc. Not the actual framework called “xUnit”).
  • Our logic extraction keeps the decision making out of the ASPX code behind. This keeps our ASPX code behind focused on the page, not buried in a code behind.
  • We have the freedom to change what this method does without affecting the page. In the future, we can make “ShouldDisplayPanelOnPage” do whatever we like.
  • Further changes to this class can be tested TDD style so we can feel confident that future refactorings won’t break functionality.

What would you have done to improve the code? Got a better idea or suggestions?
Next we’ll look at dependency injection and what happens when this decision making logic gets complex.


Kevin Jones is a Senior .NET developer and Product Team Leader at Thycotic Software Ltd. Thycotic is recognized for providing Agile TDD Training and Agile .NET Consulting Services, and its flagship password management software Secret Server.

  1. aa
    April 3, 2009 at 12:20 pm

    I would have put the list of pages in AppSettings and use a function to add the appsettings key-pair in an array/generic list.

  2. Adam Stadola
    April 3, 2009 at 6:15 pm

    @aa: I think you missed the point of Kevin’s post. The point wasn’t about WHERE to put the logic, but how to extract it and test it. Now that he put it in a separate class, he can easily unit test it.

    And for what it’s worth, AppSettings are really meant to be used as something configurable without needing a re-compile, say like on a production site. It doesn’t seem like this kind of data needs to be easily configurable.

    @Kevin, I am interested in where this is going. Is this going to be a series? A refactoring series would be sweet.

  3. Kevin Jones
    April 6, 2009 at 11:32 am

    @Adam – Yes this is going to be a series. We’ll keep taking refactoring a step further in every post. The next one should come out this coming Thursday.

  1. April 12, 2009 at 3:01 am

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: