Friday, January 22, 2010

C# Corner - fun with 'yield' and extension methods

All .NET developers learn about IEnumerable and IEnumerator in their first day of development, whether they realize it or not.  Every foreach statement written utilizes the .GetEnumerator() call to iterate over a collection of objects.  The For Each concept is a really nice piece of syntactic sugar.  However, one of its limitations is that it doesn't give you a counter to track which loop iteration you're on.  If you want to do that, you have to keep track of it yourself in your own variable.  Not a big deal really, just a fact of programming life.  This manifests itself in two ways - you can write a for;; loop which has the counter built in and then you get the object you want by indexing the collection as shown here:

var names = new List { "You", "Me", "Dupree" };

// for;; loop
Console.WriteLine("Method 1:");
for (var i = 0; i < names.Count; i++) {
   var name = names[i];
   Console.WriteLine("{0}-{1}", i, name);

This method isn't bad, but it has a fatal flaw - it only works on collections that are fully populated.  In other words, the collection has to be indexable and cannot be lazy loaded.  If you have a collection that dynamically loads or calculates its next value, you have to do something much more kludgy.  You must maintain a variable external to your loop that keeps track of your current item index as shown here:

var names = GetLazyLoadedNames();

// outside variable
Console.WriteLine("Method 2:");
var j = 0;
foreach (var name in names) {
   Console.WriteLine("{0}-{1}", j, name);

This method works on every IEnumerable collection, and is a good general purpose tactic that every .NET dev will need in their toolkit.  But, it has the frustrating consequence that it requires a variable outside the scope of the loop, and takes a couple extra lines of code to accomplish.  Both of which adversely affect the signal-to-noise ratio of your code.

As a thought experiment, I wondered if I could get the clarity of a simple and tight foreach loop while still maintaining the index.  It turns out it's really simple (in C# at least) to accomplish with a small immutable helper class and an extension method.  Here's the code:

public static class IEnumerableExtensions {
   public static IEnumerable<IndexedItem<T>> GetIndexedEnumerator<T>(this IEnumerable<T> me) {
      var index = 0;
      foreach (T item in me) {
         yield return new IndexedItem<T>(item, index);

public class IndexedItem<T> {
   private T _item;
   private int _index;

   public T Item { get { return _item; } }
   public int Index { get { return _index; } }

   public IndexedItem(T item, int index) {
      _item = item;
      _index = index;

Now, if you import the namespace containing this extension method, every IEnumerable collection you have gets a method called .GetIndexedEnumerator() which means that instead of iterating over the objects in the collection, you iterate over a new container object called IndexedItem that has the current item in the collection as well as the index of that item.  This means that now your foreach code can look like this:

var names = GetLazyLoadedNames();

// extension method
Console.WriteLine("Method 3:");
foreach (var name in names.GetIndexedEnumerator()) {
   Console.WriteLine("{0}-{1}", name.Index, name.Item);

This definitely isn't a revolutionary improvement by any means.  In fact, in exchange for clarity, you likely will see an ever-so-slightly reduced performance for very large collections since you have the added overhead of the IndexedItem class.  Also, unfortunately due to the lack of a 'yield' statement, VB devs (like myself) have a whole lot more work to do to implement this solution.  But overall, this is just one more example of why I love .NET.  The framework and languages are continually evolving, and while some purists may scoff at the syntactic sugar, I say it's all syntactic sugar over top of zeros and ones, and I have a serious sweet tooth.