Tree data structure in C#

My best advice would be that there is no standard tree data structure because there are so many ways you could implement it that it would be impossible to cover all bases with one solution. The more specific a solution, the less likely it is applicable to any given problem. I even get annoyed with LinkedList - what if I want a circular linked list?

The basic structure you'll need to implement will be a collection of nodes, and here are some options to get you started. Let's assume that the class Node is the base class of the entire solution.

If you need to only navigate down the tree, then a Node class needs a List of children.

If you need to navigate up the tree, then the Node class needs a link to its parent node.

Build an AddChild method that takes care of all the minutia of these two points and any other business logic that must be implemented (child limits, sorting the children, etc.)

delegate void TreeVisitor<T>(T nodeData);

class NTree<T>
{
    private T data;
    private LinkedList<NTree<T>> children;

    public NTree(T data)
    {
         this.data = data;
        children = new LinkedList<NTree<T>>();
    }

    public void AddChild(T data)
    {
        children.AddFirst(new NTree<T>(data));
    }

    public NTree<T> GetChild(int i)
    {
        foreach (NTree<T> n in children)
            if (--i == 0)
                return n;
        return null;
    }

    public void Traverse(NTree<T> node, TreeVisitor<T> visitor)
    {
        visitor(node.data);
        foreach (NTree<T> kid in node.children)
            Traverse(kid, visitor);
    }
}

Simple recursive implementation... < 40 lines of code... You just need to keep a reference to the root of the tree outside of the class, or wrap it in another class, maybe rename to TreeNode??

Here's mine, which is very similar to Aaron Gage's, just a little more conventional, in my opinion. For my purposes, I haven't ran into any performance issues with List<T>. It would be easy enough to switch to a LinkedList if needed.

namespace Overby.Collections
{
    public class TreeNode<T>
    {
        private readonly T _value;
        private readonly List<TreeNode<T>> _children = new List<TreeNode<T>>();

        public TreeNode(T value)
        {
            _value = value;
        }

        public TreeNode<T> this[int i]
        {
            get { return _children[i]; }
        }

        public TreeNode<T> Parent { get; private set; }

        public T Value { get { return _value; } }

        public ReadOnlyCollection<TreeNode<T>> Children
        {
            get { return _children.AsReadOnly(); }
        }

        public TreeNode<T> AddChild(T value)
        {
            var node = new TreeNode<T>(value) {Parent = this};
            _children.Add(node);
            return node;
        }

        public TreeNode<T>[] AddChildren(params T[] values)
        {
            return values.Select(AddChild).ToArray();
        }

        public bool RemoveChild(TreeNode<T> node)
        {
            return _children.Remove(node);
        }

        public void Traverse(Action<T> action)
        {
            action(Value);
            foreach (var child in _children)
                child.Traverse(action);
        }

        public IEnumerable<T> Flatten()
        {
            return new[] {Value}.Concat(_children.SelectMany(x => x.Flatten()));
        }
    }
}

Yet another tree structure:

public class TreeNode<T> : IEnumerable<TreeNode<T>>
{

    public T Data { get; set; }
    public TreeNode<T> Parent { get; set; }
    public ICollection<TreeNode<T>> Children { get; set; }

    public TreeNode(T data)
    {
        this.Data = data;
        this.Children = new LinkedList<TreeNode<T>>();
    }

    public TreeNode<T> AddChild(T child)
    {
        TreeNode<T> childNode = new TreeNode<T>(child) { Parent = this };
        this.Children.Add(childNode);
        return childNode;
    }

    ... // for iterator details see below link
}

Sample usage:

TreeNode<string> root = new TreeNode<string>("root");
{
    TreeNode<string> node0 = root.AddChild("node0");
    TreeNode<string> node1 = root.AddChild("node1");
    TreeNode<string> node2 = root.AddChild("node2");
    {
        TreeNode<string> node20 = node2.AddChild(null);
        TreeNode<string> node21 = node2.AddChild("node21");
        {
            TreeNode<string> node210 = node21.AddChild("node210");
            TreeNode<string> node211 = node21.AddChild("node211");
        }
    }
    TreeNode<string> node3 = root.AddChild("node3");
    {
        TreeNode<string> node30 = node3.AddChild("node30");
    }
}

BONUS See fully-fledged tree with:

iterator

searching

Java/C#

https://github.com/gt4dev/yet-another-tree-structure

The generally excellent C5 Generic Collection Library has several different tree-based data structures, including sets, bags and dictionaries. Source code is available if you want to study their implementation details. (I have used C5 collections in production code with good results, although I haven't used any of the tree structures specifically.)

See https://github.com/YaccConstructor/QuickGraph (previously http://quickgraph.codeplex.com/)

QuickGraph provides generic directed/undirected graph data structures and algorithms for .NET 2.0 and up. QuickGraph comes with algorithms such as depth-first search, breadth-first search, A* search, shortest path, k-shortest path, maximum flow, minimum spanning tree, least common ancestors, etc... QuickGraph supports MSAGL, GLEE, and Graphviz to render the graphs, serialization to GraphML, etc.

Here's my own:

class Program
{
    static void Main(string[] args)
    {
        var tree = new Tree<string>()
            .Begin("Fastfood")
                .Begin("Pizza")
                    .Add("Margherita")
                    .Add("Marinara")
                .End()
                .Begin("Burger")
                    .Add("Cheese burger")
                    .Add("Chili burger")
                    .Add("Rice burger")
                .End()
            .End();

        tree.Nodes.ForEach(p => PrintNode(p, 0));
        Console.ReadKey();
    }

    static void PrintNode<T>(TreeNode<T> node, int level)
    {
        Console.WriteLine("{0}{1}", new string(' ', level * 3), node.Value);
        level++;
        node.Children.ForEach(p => PrintNode(p, level));
    }
}

public class Tree<T>
{
    private Stack<TreeNode<T>> m_Stack = new Stack<TreeNode<T>>();

    public List<TreeNode<T>> Nodes { get; } = new List<TreeNode<T>>();

    public Tree<T> Begin(T val)
    {
        if (m_Stack.Count == 0)
        {
            var node = new TreeNode<T>(val, null);
            Nodes.Add(node);
            m_Stack.Push(node);
        }
        else
        {
            var node = m_Stack.Peek().Add(val);
            m_Stack.Push(node);
        }

        return this;
    }

    public Tree<T> Add(T val)
    {
        m_Stack.Peek().Add(val);
        return this;
    }

    public Tree<T> End()
    {
        m_Stack.Pop();
        return this;
    }
}

public class TreeNode<T>
{
    public T Value { get; }
    public TreeNode<T> Parent { get; }
    public List<TreeNode<T>> Children { get; }

    public TreeNode(T val, TreeNode<T> parent)
    {
        Value = val;
        Parent = parent;
        Children = new List<TreeNode<T>>();
    }

    public TreeNode<T> Add(T val)
    {
        var node = new TreeNode<T>(val, this);
        Children.Add(node);
        return node;
    }
}

Output:

Fastfood
   Pizza
      Margherita
      Marinara
   Burger
      Cheese burger
      Chili burger
      Rice burger

I have a little extension to the solutions.

Using a recursive generic declaration and a deriving subclass, you can better concentrate on your actual target.

Notice, it’s different from a non generic implementation, you don’t need to cast 'node' to 'NodeWorker'.

Here's my example:

public class GenericTree<T> where T : GenericTree<T> // recursive constraint
{
  // no specific data declaration

  protected List<T> children;

  public GenericTree()
  {
    this.children = new List<T>();
  }

  public virtual void AddChild(T newChild)
  {
    this.children.Add(newChild);
  }

  public void Traverse(Action<int, T> visitor)
  {
    this.traverse(0, visitor);
  }

  protected virtual void traverse(int depth, Action<int, T> visitor)
  {
    visitor(depth, (T)this);
    foreach (T child in this.children)
      child.traverse(depth + 1, visitor);
  }
}

public class GenericTreeNext : GenericTree<GenericTreeNext> // concrete derivation
{
  public string Name {get; set;} // user-data example

  public GenericTreeNext(string name)
  {
    this.Name = name;
  }
}

static void Main(string[] args)
{
  GenericTreeNext tree = new GenericTreeNext("Main-Harry");
  tree.AddChild(new GenericTreeNext("Main-Sub-Willy"));
  GenericTreeNext inter = new GenericTreeNext("Main-Inter-Willy");
  inter.AddChild(new GenericTreeNext("Inter-Sub-Tom"));
  inter.AddChild(new GenericTreeNext("Inter-Sub-Magda"));
  tree.AddChild(inter);
  tree.AddChild(new GenericTreeNext("Main-Sub-Chantal"));
  tree.Traverse(NodeWorker);
}

static void NodeWorker(int depth, GenericTreeNext node)
{                                // a little one-line string-concatenation (n-times)
  Console.WriteLine("{0}{1}: {2}", String.Join("   ", new string[depth + 1]), depth, node.Name);
}

Try this simple sample.

public class TreeNode<TValue>
{
    #region Properties
    public TValue Value { get; set; }
    public List<TreeNode<TValue>> Children { get; private set; }
    public bool HasChild { get { return Children.Any(); } }
    #endregion
    #region Constructor
    public TreeNode()
    {
        this.Children = new List<TreeNode<TValue>>();
    }
    public TreeNode(TValue value)
        : this()
    {
        this.Value = value;
    }
    #endregion
    #region Methods
    public void AddChild(TreeNode<TValue> treeNode)
    {
        Children.Add(treeNode);
    }
    public void AddChild(TValue value)
    {
        var treeNode = new TreeNode<TValue>(value);
        AddChild(treeNode);
    }
    #endregion
}

I created a Node<T> class that could be helpful for other people. The class has properties like:

Children

Ancestors

Descendants

Siblings

Level of the node

Parent

Root

Etc.

There is also the possibility to convert a flat list of items with an Id and a ParentId to a tree. The nodes holds a reference to both the children and the parent, so that makes iterating nodes quite fast.

There is the now released .NET codebase: specifically the code for a SortedSet that implements a red-black tree: sortedset.cs

This is, however, a balanced tree structure. So my answer is more a reference to what I believe is the only native tree-structure in the .NET core library.

I've completed the code that Berezh has shared.

public class TreeNode<T> : IEnumerable<TreeNode<T>>
{
    public T Data { get; set; }
    public TreeNode<T> Parent { get; set; }
    public ICollection<TreeNode<T>> Children { get; set; }

    public TreeNode(T data)
    {
        this.Data = data;
        this.Children = new LinkedList<TreeNode<T>>();
    }

    public TreeNode<T> AddChild(T child)
    {
        TreeNode<T> childNode = new TreeNode<T>(child) { Parent = this };
        this.Children.Add(childNode);
        return childNode;
    }

    public IEnumerator<TreeNode<T>> GetEnumerator()
    {
        throw new NotImplementedException();
    }

    IEnumerator IEnumerable.GetEnumerator()
    {
        return (IEnumerator)GetEnumerator();
    }
}

public class TreeNodeEnum<T> : IEnumerator<TreeNode<T>>
{

    int position = -1;
    public List<TreeNode<T>> Nodes { get; set; }

    public TreeNode<T> Current
    {
        get
        {
            try
            {
                return Nodes[position];
            }
            catch (IndexOutOfRangeException)
            {
                throw new InvalidOperationException();
            }
        }
    }

    object IEnumerator.Current
    {
        get
        {
            return Current;
        }
    }

    public TreeNodeEnum(List<TreeNode<T>> nodes)
    {
        Nodes = nodes;
    }

    public void Dispose()
    {

    }

    public bool MoveNext()
    {
        position++;
        return (position < Nodes.Count);
    }

    public void Reset()
    {
        position = -1;
    }
}

Most trees are formed by the data you are processing.

Say you have a person class that includes details of someone’s parents, would you rather have the tree structure as part of your “domain class”, or use a separate tree class that contained links to your person objects? Think about a simple operation like getting all the grandchildren of a person, should this code be in the person class, or should the user of the person class have to know about a separate tree class?

Another example is a parse tree in a compiler…

Both of these examples show that the concept of a tree is part of the domain of the data and using a separate general-purpose tree at least doubles the number of objects that are created as well as making the API harder to program again.

We want a way to reuse the standard tree operations, without having to reimplement them for all trees, while at the same time, not having to use a standard tree class. Boost has tried to solve this type of problem for C++, but I am yet to see any effect for .NET to get it adapted.

I have added a complete solution and example using the NTree class above. I also added the "AddChild" method...

    public class NTree<T>
    {
        public T data;
        public LinkedList<NTree<T>> children;

        public NTree(T data)
        {
            this.data = data;
            children = new LinkedList<NTree<T>>();
        }

        public void AddChild(T data)
        {
            var node = new NTree<T>(data) { Parent = this };
            children.AddFirst(node);
        }

        public NTree<T> Parent { get; private set; }

        public NTree<T> GetChild(int i)
        {
            foreach (NTree<T> n in children)
                if (--i == 0)
                    return n;
            return null;
        }

        public void Traverse(NTree<T> node, TreeVisitor<T> visitor, string t, ref NTree<T> r)
        {
            visitor(node.data, node, t, ref r);
            foreach (NTree<T> kid in node.children)
                Traverse(kid, visitor, t, ref r);
        }
    }

    public static void DelegateMethod(KeyValuePair<string, string> data, NTree<KeyValuePair<string, string>> node, string t, ref NTree<KeyValuePair<string, string>> r)
    {
        string a = string.Empty;
        if (node.data.Key == t)
        {
            r = node;
            return;
        }
    }

Using it

 NTree<KeyValuePair<string, string>> ret = null;
 tree.Traverse(tree, DelegateMethod, node["categoryId"].InnerText, ref ret);

There is also the possibility to use XML with LINQ:

Create XML tree in C# (LINQ to XML)

XML is the most mature and flexible solution when it comes to using trees and LINQ provides you with all the tools that you need. The configuration of your tree also gets much cleaner and user-friendly as you can simply use an XML file for the initialization.

If you need to work with objects, you can use XML serialization:

XML serialization

If you are going to display this tree on the GUI, you can use TreeView and TreeNode. (I suppose technically you can create a TreeNode without putting it on a GUI, but it does have more overhead than a simple homegrown TreeNode implementation.)

Here is my implementation of a BST:

class BST
{
    public class Node
    {
        public Node Left { get; set; }
        public object Data { get; set; }
        public Node Right { get; set; }

        public Node()
        {
            Data = null;
        }

        public Node(int Data)
        {
            this.Data = (object)Data;
        }

        public void Insert(int Data)
        {
            if (this.Data == null)
            {
                this.Data = (object)Data;
                return;
            }
            if (Data > (int)this.Data)
            {
                if (this.Right == null)
                {
                    this.Right = new Node(Data);
                }
                else
                {
                    this.Right.Insert(Data);
                }
            }
            if (Data <= (int)this.Data)
            {
                if (this.Left == null)
                {
                    this.Left = new Node(Data);
                }
                else
                {
                    this.Left.Insert(Data);
                }
            }
        }

        public void TraverseInOrder()
        {
            if(this.Left != null)
                this.Left.TraverseInOrder();
            Console.Write("{0} ", this.Data);
            if (this.Right != null)
                this.Right.TraverseInOrder();
        }
    }

    public Node Root { get; set; }
    public BST()
    {
        Root = new Node();
    }
}

Tree With Generic Data

using System;
using System.Collections.Concurrent;
using System.Collections.Generic;
using System.Linq;
using System.Threading;
using System.Threading.Tasks;

public class Tree<T>
{
    public T Data { get; set; }
    public LinkedList<Tree<T>> Children { get; set; } = new LinkedList<Tree<T>>();
    public Task Traverse(Func<T, Task> actionOnNode, int maxDegreeOfParallelism = 1) => Traverse(actionOnNode, new SemaphoreSlim(maxDegreeOfParallelism, maxDegreeOfParallelism));
    private async Task Traverse(Func<T, Task> actionOnNode, SemaphoreSlim semaphore)
    {
        await actionOnNode(Data);
        SafeRelease(semaphore);
        IEnumerable<Task> tasks = Children.Select(async input =>
        {
            await semaphore.WaitAsync().ConfigureAwait(false);
            try
            {
                await input.Traverse(actionOnNode, semaphore).ConfigureAwait(false);
            }
            finally
            {
                SafeRelease(semaphore);
            }
        });
        await Task.WhenAll(tasks);
    }
    private void SafeRelease(SemaphoreSlim semaphore)
    {
        try
        {
            semaphore.Release();
        }
        catch (Exception ex)
        {
            if (ex.Message.ToLower() != "Adding the specified count to the semaphore would cause it to exceed its maximum count.".ToLower())
            {
                throw;
            }
        }
    }

    public async Task<IEnumerable<T>> ToList()
    {
        ConcurrentBag<T> lst = new ConcurrentBag<T>();
        await Traverse(async (data) => lst.Add(data));
        return lst;
    }
    public async Task<int> Count() => (await ToList()).Count();
}

Unit Tests

using System.Threading.Tasks;
using Xunit;

public class Tree_Tests
{
    [Fact]
    public async Task Tree_ToList_Count()
    {
        Tree<int> head = new Tree<int>();

        Assert.NotEmpty(await head.ToList());
        Assert.True(await head.Count() == 1);

        // child
        var child = new Tree<int>();
        head.Children.AddFirst(child);
        Assert.True(await head.Count() == 2);
        Assert.NotEmpty(await head.ToList());

        // grandson
        child.Children.AddFirst(new Tree<int>());
        child.Children.AddFirst(new Tree<int>());
        Assert.True(await head.Count() == 4);
        Assert.NotEmpty(await head.ToList());
    }

    [Fact]
    public async Task Tree_Traverse()
    {
        Tree<int> head = new Tree<int>() { Data = 1 };

        // child
        var child = new Tree<int>() { Data = 2 };
        head.Children.AddFirst(child);

        // grandson
        child.Children.AddFirst(new Tree<int>() { Data = 3 });
        child.Children.AddLast(new Tree<int>() { Data = 4 });

        int counter = 0;
        await head.Traverse(async (data) => counter += data);
        Assert.True(counter == 10);

        counter = 0;
        await child.Traverse(async (data) => counter += data);
        Assert.True(counter == 9);

        counter = 0;
        await child.Children.First!.Value.Traverse(async (data) => counter += data);
        Assert.True(counter == 3);

        counter = 0;
        await child.Children.Last!.Value.Traverse(async (data) => counter += data);
        Assert.True(counter == 4);
    }
}

I don't like a tree aproach. It gets things overcomplicated including search or dril-down or even ui controls populating.

I would suggest to use a very simple approach with IDictionary<TChild, TParent>. This also allows to have no connections between nodes or levels.

In case you need a rooted tree data structure implementation that uses less memory, you can write your Node class as follows (C++ implementation):

class Node {
       Node* parent;
       int item; // depending on your needs

       Node* firstChild; //pointer to left most child of node
       Node* nextSibling; //pointer to the sibling to the right
}