C# algorithm for generating hierarchy

Question

I've got a text file that looks like this:

{ Id = 1, ParentId = 0, Position = 0, Title = "root" }
{ Id = 2, ParentId = 1, Position = 0, Title = "child 1" }
{ Id = 3, ParentId = 1, Position = 1, Title = "child 2" }
{ Id = 4, ParentId = 1, Position = 2, Title = "child 3" }
{ Id = 5, ParentId = 4, Position = 0, Title = "grandchild 1" }

I'm looking for a generic C# algorithm that will create an object hierarchy from this. A "Hierarchize" function, if you will, that turns this data into an object hierarchy.

Any ideas?

edit I've already parsed the file into .NET objects:

class Node
{
    public int Id { get; }
    public int ParentId { get; }
    public int Position { get; }
    public string Title { get; }
}

Now I need to actually arrange the objects into an object graph.

Answer 1

Are you certain the last line's ParentID is 1? The title says grandchild, but it would be a child of "root" if i'm reading things correctly.

Answer 2

Hmm... I don't quite see how that works. How can 2 and 5 both have parent=1, position=0? Should 5 have parent 2, 3 or 4?

Okay, this new version goes through the all the nodes three times:

• Load all nodes and put them into the a map
• Associate each node with its parent
• Sort each node's child by position

It's not well-encapsulated, nicely error checking etc - but it works.

using System;
using System.Collections.Generic;
using System.IO;

public class Node
{
    private static readonly char[] Braces = "{}".ToCharArray();
    private static readonly char[] StringTrim = "\" ".ToCharArray();

    public Node Parent { get; set; }
    public int ParentId { get; private set; }
    public int Id { get; private set; }
    public string Title { get; private set; }
    public int Position { get; private set; }
    private readonly List<Node> children = new List<Node>();
    public List<Node> Children { get { return children; } }

    public static Node FromLine(string line)
    {
        Node node = new Node();
        line = line.Trim(Braces);
        string[] bits = line.Split(',');
        foreach (string bit in bits)
        {
            string[] keyValue = bit.Split('=');
            string key = keyValue[0].Trim();
            string value = keyValue[1].Trim();
            switch (key)
            {
                case "Id":
                    node.Id = int.Parse(value);
                    break;
                case "ParentId":
                    node.ParentId = int.Parse(value);
                    break;
                case "Position":
                    node.Position = int.Parse(value);
                    break;
                case "Title":
                    node.Title = value.Trim(StringTrim);
                    break;
                default:
                    throw new ArgumentException("Bad line: " + line);
            }
        }
        return node;
    }

    public void Dump()
    {
        int depth = 0;
        Node node = this;
        while (node.Parent != null)
        {
            depth++;
            node = node.Parent;
        }
        Console.WriteLine(new string(' ', depth * 2) + Title);
        foreach (Node child in Children)
        {
            child.Dump();
        }
    }
}

class Test
{       
    static void Main(string[] args)
    {
        var dictionary = new Dictionary<int, Node>();

        using (TextReader reader = File.OpenText("test.txt"))
        {
            string line;
            while ((line = reader.ReadLine()) != null)
            {
                Node node = Node.FromLine(line);
                dictionary[node.Id] = node;
            }
        }
        foreach (Node node in dictionary.Values)
        {
            if (node.ParentId != 0)
            {
                node.Parent = dictionary[node.ParentId];
                node.Parent.Children.Add(node);
            }
        }

        foreach (Node node in dictionary.Values)
        {
            node.Children.Sort((n1, n2) =>
                               n1.Position.CompareTo(n2.Position));
        }

        Node root = dictionary[1];
        root.Dump();
    }
}

Sample text file:

{ Id = 5, ParentId = 4, Position = 0, Title = "grandchild 1" }
{ Id = 2, ParentId = 1, Position = 0, Title = "child 1" }
{ Id = 4, ParentId = 1, Position = 2, Title = "child 3" }
{ Id = 3, ParentId = 1, Position = 1, Title = "child 2" }
{ Id = 1, ParentId = 0, Position = 0, Title = "root" }

Output:

root
  child 1
  child 2
  child 3
    grandchild 1

Answer 3

Once you have the file parsed in you can follow this blog on how to assemble the objects into a hierarchy using LINQ.

Answer 4

class Node {
    public int Id { get;set;  }
    public int ParentId { get;set;  }
    public int Position { get;set;  }
    public string Title { get;set;  }
    public IEnumerable<Node> Children { get; set; }

    public override string ToString() { return ToString(0); }
    public string ToString(int depth) {
        return "\n" + new string(' ', depth * 2) + Title + (
            Children.Count() == 0 ? "" :
            string.Join("", Children
                .Select(node => node.ToString(depth + 1))
                .ToArray()
            );
    }
}
class Program {
    static void Main(string[] args) {
        var data = new[] {
            new Node{ Id = 1, ParentId = 0, Position = 0, Title = "root" },
            new Node{ Id = 2, ParentId = 1, Position = 0, Title = "child 1" },
            new Node{ Id = 3, ParentId = 1, Position = 1, Title = "child 2" },
            new Node{ Id = 4, ParentId = 1, Position = 2, Title = "child 3" },
            new Node{ Id = 5, ParentId = 3, Position = 0, Title = "grandchild 1" }
        };
        Func<Node, Node> transform = null;
        transform = node => new Node {
            Title = node.Title,
            Id = node.Id,
            ParentId = node.ParentId,
            Position = node.Position,
            Children = (
                from child in data
                where child.ParentId == node.Id
                orderby child.Position
                select transform(child))
        };
        Console.WriteLine(transform(data[0]));
    }
}

result:

root
  child 1
  child 2
    grandchild 1
  child 3

Answer 5

I assume that your example incorrectly gives the wrong parent ID to object #5. This should cover it. Caveats: Assumes the "topmost" node always has a parent ID of zero. Ignores any nodes that aren't eventually descended from the topmost node. Behavior will be odd if presented with duplicate IDs.

public class FlatObj
{
    public int Id;
    public int ParentId;
    public int Position;
    public string Title;
}

public class Node
{
    public int ID;
    public string Title;
    public IList<Node> Children;

    public Node(FlatObject baseObject, IList<Node> children)
    {
        this.ID = baseObject.Id;
        this.Title = baseObject.Title;
        this.Children = children;
    }
}

public static Node CreateHierarchy(IEnumerable<FlatObject> objects)
{
    var families = objects.ToLookup(x => x.ParentId);
    var topmost = families[0].Single();

    Func<int, IList<Node>> Children = null;

    Children = (parentID) => families[parentID]
        .OrderBy(x => x.Position)
        .Select(x => new Node(x, Children(x.Id))).ToList();

    return new Node(topmost, Children(topmost.Id));
}

public static void Test()
{
    List<FlatObj> objects = new List<FlatObj> {
    new FlatObj { Id = 1, ParentId = 0, Position = 0, Title = "root" },
    new FlatObj { Id = 2, ParentId = 1, Position = 0, Title = "child 1" },
    new FlatObj { Id = 3, ParentId = 1, Position = 1, Title = "child 2" },
    new FlatObj { Id = 4, ParentId = 1, Position = 2, Title = "child 3" },
    new FlatObj { Id = 5, ParentId = 2, Position = 0, Title = "grandchild" }};

    var nodes = CreateHierarchy(objects);
}

Answer 6

Many thanks to Jon and to mquander - you guys gave me enough information to help me solve this in a proper, generic way. Here's my solution, a single generic extension method that converts objects into hierarchy form:

public static IEnumerable<Node<T>> Hierarchize<T, TKey, TOrderKey>(
    this IEnumerable<T> elements, 
    TKey topMostKey, 
    Func<T, TKey> keySelector, 
    Func<T, TKey> parentKeySelector, 
    Func<T, TOrderKey> orderingKeySelector)
{
    var families = elements.ToLookup(parentKeySelector);
    Func<TKey, IEnumerable<Node<T>>> Children = null;
    Children = parentId => families[parentId]
        .OrderBy(orderingKeySelector)
        .Select(x => new Node<T>(x, Children(keySelector(x))));

    return Children(topMostKey);
}

Utilizes this small data node class:

public class Node<T>
{
    public T DataObject { get; private set; }
    public IList<Node<T>> Children { get; private set; }

    public Node(T dataObject, IEnumerable<Node<T>> children)
    {
        this.DataObject = dataObject;
        this.Children = new List<Node<T>>(children);
    }
}

It's generic enough to work for a variety of problems, including my text file issue. Nifty!

Answer 7

Judah and all, thank you all for ending up such a generic way to get hierarchy structure which makes my life much easier. I am wonder if you can help me to implement a little bit more by adding a level property (aka depth). The code by Omer van (http://weblogs.asp.net/okloeten/archive/2006/07/09/Hierarchical-Linq-Queries.aspx) has level property but I do not know how to integrate two methods together.