Sort ObservableCollection through C#

Question

I have below ObservableCollection<string>. I need to sort this alphabetically.

private ObservableCollection<string> _animals = new ObservableCollection<string>
{
    "Cat", "Dog", "Bear", "Lion", "Mouse",
    "Horse", "Rat", "Elephant", "Kangaroo", "Lizard", 
    "Snake", "Frog", "Fish", "Butterfly", "Human", 
    "Cow", "Bumble Bee"
};

I tried _animals.OrderByDescending. But I don't know how to use it correctly.

_animals.OrderByDescending(a => a.<what_is_here_?>);

How can I do this?

Answer 1

Introduction

Basically, if there is a need to display a sorted collection, please consider using the CollectionViewSource class: assign ("bind") its Source property to the source collection — an instance of the ObservableCollection<T> class.

The idea is that CollectionViewSource class provides an instance of the CollectionView class. This is kind of "projection" of the original (source) collection, but with applied sorting, filtering, etc.

References:

• How to: Sort and Group Data Using a View in XAML.
• WPF’s CollectionViewSource.

Live Shaping

WPF 4.5 introduces "Live Shaping" feature for CollectionViewSource.

References:

• WPF 4.5 New Feature: Live Shaping.
• CollectionViewSource.IsLiveSorting Property.
• Repositioning data as the data's values change (Live shaping).

Solution

If there still a need to sort an instance of the ObservableCollection<T> class, here is how it can be done. The ObservableCollection<T> class itself does not have sort method. But, the collection could be re-created to have items sorted:

// Animals property setter must raise "property changed" event to notify binding clients.
// See INotifyPropertyChanged interface for details.
Animals = new ObservableCollection<string>
    {
        "Cat", "Dog", "Bear", "Lion", "Mouse",
        "Horse", "Rat", "Elephant", "Kangaroo",
        "Lizard", "Snake", "Frog", "Fish",
        "Butterfly", "Human", "Cow", "Bumble Bee"
    };
...
Animals = new ObservableCollection<string>(Animals.OrderBy(i => i));

Additional details

Please note that OrderBy() and OrderByDescending() methods (as other LINQ–extension methods) do not modify the source collection! They instead create a new sequence (i.e. a new instance of the class that implements IEnumerable<T> interface). Thus, it is necessary to re-create the collection.

Answer 2

The way

The way I would go is to build a List<> starting from the ObservableCollection<>, sort it (through its Sort() method, more on msdn) and when the List<> has been sorted, reorder the ObservableCollection<> with the Move() method.

The code

public static void Sort<T>(this ObservableCollection<T> collection, Comparison<T> comparison)
{
    var sortableList = new List<T>(collection);
    sortableList.Sort(comparison);

    for (int i = 0; i < sortableList.Count; i++)
    {
        collection.Move(collection.IndexOf(sortableList[i]), i);
    }
}

The test

public void TestObservableCollectionSortExtension()
{
    var observableCollection = new ObservableCollection<int>();
    var maxValue = 10;

    // Populate the list in reverse mode [maxValue, maxValue-1, ..., 1, 0]
    for (int i = maxValue; i >= 0; i--)
    {
        observableCollection.Add(i);
    }

    // Assert the collection is in reverse mode
    for (int i = maxValue; i >= 0; i--)
    {
        Assert.AreEqual(i, observableCollection[maxValue - i]);
    }

    // Sort the observable collection
    observableCollection.Sort((a, b) => { return a.CompareTo(b); });

    // Assert elements have been sorted
    for (int i = 0; i < maxValue; i++)
    {
        Assert.AreEqual(i, observableCollection[i]);
    }
}

##Notes This is just a proof of concept, showing how to sort an ObservableCollection<> without breaking the bindings on items.The sort algorithm has room for improvements and validations (like index checking as pointed out here).

Answer 3

I looked at these, I was getting it sorted, and then it broke the binding, as above. Came up with this solution, though simpler than most of yours, it appears to do what I want to,,,

public static ObservableCollection<string> OrderThoseGroups( ObservableCollection<string> orderThoseGroups)
    {
        ObservableCollection<string> temp;
        temp =  new ObservableCollection<string>(orderThoseGroups.OrderBy(p => p));
        orderThoseGroups.Clear();
        foreach (string j in temp) orderThoseGroups.Add(j);
        return orderThoseGroups;



    }

Answer 4

This extension method eliminates the need to sort the entire list.

Instead, it inserts each new item in place. So the list is always remains sorted.

It turns out that this method just works when a lot of the other methods fail due to missing notifications when the collection changes. And it is rather fast.

The code below should be bulletproof; it has been extensively tested in a large-scale production environment.

To use:

// Call on dispatcher.
ObservableCollection<MyClass> collectionView = new ObservableCollection<MyClass>();
var p1 = new MyClass() { Key = "A" }
var p2 = new MyClass() { Key = "Z" }
var p3 = new MyClass() { Key = "D" }
collectionView.InsertInPlace(p1, o => o.Key);
collectionView.InsertInPlace(p2, o => o.Key);
collectionView.InsertInPlace(p3, o => o.Key);
// The list will always remain ordered on the screen, e.g. "A, D, Z" .
// Insertion speed is Log(N) as it uses a binary search.

And the extension method:

/// <summary>
/// Inserts an item into a list in the correct place, based on the provided key and key comparer. Use like OrderBy(o => o.PropertyWithKey).
/// </summary>
public static void InsertInPlace<TItem, TKey>(this ObservableCollection<TItem> collection, TItem itemToAdd, Func<TItem, TKey> keyGetter)
{
    int index = collection.ToList().BinarySearch(keyGetter(itemToAdd), Comparer<TKey>.Default, keyGetter);
    collection.Insert(index, itemToAdd);
}

And the binary search extension method:

/// <summary>
/// Binary search.
/// </summary>
/// <returns>Index of item in collection.</returns> 
/// <notes>This version tops out at approximately 25% faster than the equivalent recursive version. This 25% speedup is for list
/// lengths more of than 1000 items, with less performance advantage for smaller lists.</notes>
public static int BinarySearch<TItem, TKey>(this IList<TItem> collection, TKey keyToFind, IComparer<TKey> comparer, Func<TItem, TKey> keyGetter)
{
    if (collection == null)
    {
        throw new ArgumentNullException(nameof(collection));
    }

    int lower = 0;
    int upper = collection.Count - 1;

    while (lower <= upper)
    {
        int middle = lower + (upper - lower) / 2;
        int comparisonResult = comparer.Compare(keyToFind, keyGetter.Invoke(collection[middle]));
        if (comparisonResult == 0)
        {
            return middle;
        }
        else if (comparisonResult < 0)
        {
            upper = middle - 1;
        }
        else
        {
            lower = middle + 1;
        }
    }

    // If we cannot find the item, return the item below it, so the new item will be inserted next.
    return lower;
}

Answer 5

This is an ObservableCollection<T>, that automatically sorts itself upon a change, triggers a sort only when necessary, and only triggers a single move collection change action.

using System;
using System.Collections.Generic;
using System.Collections.ObjectModel;
using System.Collections.Specialized;
using System.Linq;

namespace ConsoleApp4
{
  using static Console;

  public class SortableObservableCollection<T> : ObservableCollection<T>
  {
    public Func<T, object> SortingSelector { get; set; }
    public bool Descending { get; set; }
    protected override void OnCollectionChanged(NotifyCollectionChangedEventArgs e)
    {
      base.OnCollectionChanged(e);
      if (SortingSelector == null 
          || e.Action == NotifyCollectionChangedAction.Remove
          || e.Action == NotifyCollectionChangedAction.Reset)
        return;
                      
      var query = this
        .Select((item, index) => (Item: item, Index: index));
      query = Descending
        ? query.OrderByDescending(tuple => SortingSelector(tuple.Item))
        : query.OrderBy(tuple => SortingSelector(tuple.Item));

      var map = query.Select((tuple, index) => (OldIndex:tuple.Index, NewIndex:index))
       .Where(o => o.OldIndex != o.NewIndex);

      using (var enumerator = map.GetEnumerator())
       if (enumerator.MoveNext())
          Move(enumerator.Current.OldIndex, enumerator.Current.NewIndex);


    }
  }

  
  //USAGE
  class Program
  {
    static void Main(string[] args)
    {
      var xx = new SortableObservableCollection<int>() { SortingSelector = i => i };
      xx.CollectionChanged += (sender, e) =>
       WriteLine($"action: {e.Action}, oldIndex:{e.OldStartingIndex},"
         + " newIndex:{e.NewStartingIndex}, newValue: {xx[e.NewStartingIndex]}");

      xx.Add(10);
      xx.Add(8);
      xx.Add(45);
      xx.Add(0);
      xx.Add(100);
      xx.Add(-800);
      xx.Add(4857);
      xx.Add(-1);

      foreach (var item in xx)
        Write($"{item}, ");
    }
  }
}

Output:

action: Add, oldIndex:-1, newIndex:0, newValue: 10
action: Add, oldIndex:-1, newIndex:1, newValue: 8
action: Move, oldIndex:1, newIndex:0, newValue: 8
action: Add, oldIndex:-1, newIndex:2, newValue: 45
action: Add, oldIndex:-1, newIndex:3, newValue: 0
action: Move, oldIndex:3, newIndex:0, newValue: 0
action: Add, oldIndex:-1, newIndex:4, newValue: 100
action: Add, oldIndex:-1, newIndex:5, newValue: -800
action: Move, oldIndex:5, newIndex:0, newValue: -800
action: Add, oldIndex:-1, newIndex:6, newValue: 4857
action: Add, oldIndex:-1, newIndex:7, newValue: -1
action: Move, oldIndex:7, newIndex:1, newValue: -1
-800, -1, 0, 8, 10, 45, 100, 4857,

Answer 6

I created an extension method to the ObservableCollection

public static void MySort<TSource,TKey>(this ObservableCollection<TSource> observableCollection, Func<TSource, TKey> keySelector)
    {
        var a = observableCollection.OrderBy(keySelector).ToList();
        observableCollection.Clear();
        foreach(var b in a)
        {
            observableCollection.Add(b);
        }
    }

It seems to work and you don't need to implement IComparable

Answer 7

The argument to OrderByDescending is a function returning a key to sort with. In your case, the key is the string itself:

var result = _animals.OrderByDescending(a => a);

If you wanted to sort by length for example, you'll write:

var result = _animals.OrderByDescending(a => a.Length);

Answer 8

myObservableCollection.ToList().Sort((x, y) => x.Property.CompareTo(y.Property));

Answer 9

_animals.OrderByDescending(a => a.<what_is_here_?>);

If animals would be a list of object Animal, you could use a property to order the list.

public class Animal
{
    public int ID {get; set;}
    public string Name {get; set;}
    ...
}

ObservableCollection<Animal> animals = ...
animals = animals.OrderByDescending(a => a.Name);

Answer 10

/// <summary>
/// Sorts the collection.
/// </summary>
/// <typeparam name="T">The type of the elements of the collection.</typeparam>
/// <param name="collection">The collection to sort.</param>
/// <param name="comparison">The comparison used for sorting.</param>
public static void Sort<T>(this ObservableCollection<T> collection, Comparison<T> comparison = null)
{
    var sortableList = new List<T>(collection);
    if (comparison == null)
        sortableList.Sort();
    else
        sortableList.Sort(comparison);

    for (var i = 0; i < sortableList.Count; i++)
    {
        var oldIndex = collection.IndexOf(sortableList[i]);
        var newIndex = i;
        if (oldIndex != newIndex)
            collection.Move(oldIndex, newIndex);
    }
}

This solution is based on Marco's answer. I had some problems with his solution and therefore improved it by only calling Move if the index actually changed. This should improve performance and also fix the linked issue.

Answer 11

I would like to share my thoughts as well, since I've bumped into the same issue.

Well, just answering the question would be:

1 - Add an extenssion to the observable collection class like this:

namespace YourNameSpace
{
    public static class ObservableCollectionExtension
    {
        public static void OrderByReference<T>(this ObservableCollection<T> collection, List<T> comparison)
        {
            for (int i = 0; i < comparison.Count; i++)
            {
                if (!comparison.ElementAt(i).Equals(collection.ElementAt(i)))
                    collection.Move(collection.IndexOf(comparison[i]), i);
            }
        }
        
        public static void InsertInPlace<T>(this ObservableCollection<T> collection, List<T> comparison, T item)
        {
            int index = comparison.IndexOf(item);
            comparison.RemoveAt(index);
            collection.OrderByReference(comparison);
            collection.Insert(index, item);
        }
    }
}

2 - Then use it like this:

_animals.OrderByReference(_animals.OrderBy(x => x).ToList());

This changes your ObservableCollection, you can use linq and it doesn't change the bindings!

Extra:

I've extended @Marco and @Contango answers to my own liking. First I thought of using a list directly as the comparison, so you would have this:

public static void OrderByReference<T>(this ObservableCollection<T> collection, List<T> comparison)
{
    for (int i = 0; i < comparison.Count; i++)
    {
        collection.Move(collection.IndexOf(comparison[i]), i);
    }
}

And using like this:

YourObservableCollection.OrderByReference(YourObservableCollection.DoYourLinqOrdering().ToList());

Then I've thought, since this always move everything and triggers the move in the ObservableCollection why not compare if the object is already in there, and this brings what I've put in the begining with the Equals comparator.

Adding the object to the correct place also sounded good, but I wanned a simple way to do it. So I've came up with that:

public static void InsertInPlace<T>(this ObservableCollection<T> collection, List<T> comparison, T item)
{
    collection.Insert(comparison.IndexOf(item), item);
}

You send a list with the new object where you want and also this new object, so you need to create a list, then add this new object, like this:

var YourList = YourObservableCollection.ToList();
var YourObject = new YourClass { ..... };
YourList.Add(YourObject);
YourObservableCollection.InsertInPlace(YourList.DoYourLinqOrdering().ToList(), YourObject);

But since the ObservableCollection could be in a different order than the list because of the selection in the "DoYourLinqOrdering()" (this would happen if the collection wasn't previously ordered) I've added the first extession (OrderByReference) in the insert as you can see in the begining of the answer. It will not take long if it doesn't need to move the itens arround, so I did't saw a problem in using it.

As performance goes, I've compared the methods by checking the time it takes for each to finish, so not ideal, but anyway, I've tested an observable collection with 20000 itens. For the OrderByReference I didn't saw great difference in the performance by adding the Equal object checker, but if not all itens need to be moved it is faster and it doesn't fire unecessary Move events on the collecitonChanged, so thats something. For the InsertInPlace is the same thing, if the ObservableCollection is already sorted, just checking if the objects are in the right place is faster than moving all the itens around, so there was not a huge difference in time if it is just passing through the Equals statement and you get the benefit of being sure everything is where it should be.

Be aware that if you use this extession with objects that dont mach or with a list that have more or less objects you will get an ArgumentOutOfRangeException or some other unexpect behaviour.

Hopes this helps somebody!

Answer 12

I did a sort on a certain class field (distance).

public class RateInfo 
{
    public string begin { get; set; }
    public string end { get; set; }
    public string price { get; set; }
    public string comment { get; set; }
    public string phone { get; set; }
    public string ImagePath { get; set; }
    public string what { get; set; }
    public string distance { get; set; }
}    

public ObservableCollection<RateInfo> Phones { get; set; }

public List<RateInfo> LRate { get; set; }

public ObservableCollection<RateInfo> Phones { get; set; }

public List<RateInfo> LRate { get; set; }

......

foreach (var item in ph)
        {

            LRate.Add(new RateInfo { begin = item["begin"].ToString(), end = item["end"].ToString(), price = item["price"].ToString(), distance=kilom, ImagePath = "chel.png" });
        }

       LRate.Sort((x, y) => x.distance.CompareTo(y.distance));

        foreach (var item in LRate)
        {
            Phones.Add(item);
        }

Answer 13

Here is a slight variation on Shimmy's one for collection of classes that already implement the well-known IComparable<T> interface. In this case, the "order by" selector is implicit.

public class SortedObservableCollection<T> : ObservableCollection<T> where T : IComparable<T>
{
    protected override void OnCollectionChanged(NotifyCollectionChangedEventArgs e)
    {
        base.OnCollectionChanged(e);
        if (e.Action != NotifyCollectionChangedAction.Reset &&
            e.Action != NotifyCollectionChangedAction.Move &&
            e.Action != NotifyCollectionChangedAction.Remove)
        {
            var query = this.Select((item, index) => (Item: item, Index: index)).OrderBy(tuple => tuple.Item, Comparer.Default);
            var map = query.Select((tuple, index) => (OldIndex: tuple.Index, NewIndex: index)).Where(o => o.OldIndex != o.NewIndex);
            using (var enumerator = map.GetEnumerator())
            {
                if (enumerator.MoveNext())
                {
                    base.MoveItem(enumerator.Current.OldIndex, enumerator.Current.NewIndex);
                }
            }
        }
    }

    // (optional) user is not allowed to move items in a sorted collection
    protected override void MoveItem(int oldIndex, int newIndex) => throw new InvalidOperationException();
    protected override void SetItem(int index, T item) => throw new InvalidOperationException();

    private class Comparer : IComparer<T>
    {
        public static readonly Comparer Default = new Comparer();

        public int Compare(T x, T y) => x.CompareTo(y);
    }

    // explicit sort; sometimes needed.
    public virtual void Sort()
    {
        if (Items.Count <= 1)
            return;

        var items = Items.ToList();
        Items.Clear();
        items.Sort();
        foreach (var item in items)
        {
            Items.Add(item);
        }
        OnPropertyChanged(new PropertyChangedEventArgs("Item[]"));
        OnCollectionChanged(new NotifyCollectionChangedEventArgs(NotifyCollectionChangedAction.Reset));
    }
}

Answer 14

If performance is your main concern and you dont mind listening to different events, then this is the way to go for a stable sort:

public static void Sort<T>(this ObservableCollection<T> list) where T : IComparable<T>
{
    int i = 0;
    foreach (var item in list.OrderBy(x => x))
    {
        if (!item.Equals(list[i]))
        {
            list[i] = item;
        }

        i++;
    }
}

I am not sure if there is anything simpler and faster (at least theoretically), as far as stable sorts go. Doing a ToArray on the ordered list might make the enumeration faster but at worse space complexity. You could also do away with the Equals check to go even faster, but I guess reducing change notification is a welcome thing.

Also this doesn't break any bindings.

Mind you this raises a bunch of Replace events rather than Move (which is more expected for a Sort operation), and also the number of events raised will be most likely more when compared to other Move approaches in this thread but it is unlikely it matters for performance, I think.. Most UI elements must have implemented IList and doing a replace on ILists should be faster than Moves. But more changed events means more screen refreshes. You will have to test it out to see the implications.

---

For a Move answer, see this. Haven't seen a more correct implementation that works even when you have duplicates in the collection.

Answer 15

    public static ObservableCollection<T> Sort<T>(this ObservableCollection<T> collection, Comparison<T> comparison)
    {
        var sortableList = new List<T>(collection);
        sortableList.Sort(comparison);
        collection.Clear();
        sortableList.ForEach(item => { collection.Add(item); });
        return collection;
    }