Develop Reference

How would I search through all nested children of a custom class?

I have a table in a database which looks like this:
| id | parentID | name |
|----------+----------+-------------|
|ABCD-12345| | Top |
|----------+----------+-------------|
|ABCD-23456|ABCD-12345| Middle |
|----------+----------+-------------|
|ABCD-34567|ABCD-23456| Bottom |
|----------+----------+-------------|
|ABCD-45678|ABCD-23456| Bottom |
etc. - Basically, a hierarchical structure of N depth. I've taken this and shoved it into a datatable.
I have the following class built to hold this data:
public class TreeNode
{
public string id { get; set; }
public string name { get; set; }
public string parentID { get; set; }
public List<TreeNode> children { get; set; }
}
My goal is to go through each of these DataTable rows and insert them into the appropriate location in the TreeNode structure, but I'm super confused as to how I should approach this.
The main point of confusion for me is how I search through the entire existing structure of TreeNodes to see if a node with the parentID exists. Can anyone point me in the right direction?
I tried the following code, but it doesn't work:
public List<TreeNode> BuildTree(int currNode, List<TreeNode> treeList, DataTable dt)
{
foreach(DataRow row in dt.Rows)
{
if(row[1].ToString() == treeList[currNode].id)
{
treeList[currNode].children.Add(new TreeNode
{
id = row[0].ToString(),
name = row[2].ToString(),
parentID = row[1].ToString()
});
dt.Rows.Remove(row);
if(dt.Rows.Count > 0)
{
currNode++;
BuildTree(currNode, treeList, dt);
}
else
{
return treeList;
}
}
}
return null;
}
The problem is this line:
if(row[1].ToString() == treeList[currNode].id)
which gets an out of range exception, because I have a root at index 0, so on the second run (when currNode is 1), it breaks. I need to traverse to treeList[0].Children[int], followed by treeList[0].Children[int].Children[int] and so on and so forth.
So how do I accomplish this goal?

First I'm going to modify the TreeNode class for our convenience. It's not necessary, but just a nice to have. Also I'm going to assume that in your datatable you've done your error checking and there's only one node with ParentId = "".
public class TreeNode
{
public string Id { get; set; }
public string Name { get; set; }
public string ParentID { get; set; }
public List<TreeNode> Children { get; set; }
public TreeNode()
{
Id = Name = ParentID = string.Empty;
Children = new List<TreeNode>();
}
public bool IsRoot { get { return ParentID == string.Empty; } }
public bool IsChild { get { return Children == null || Children.Count == 0; } }
}
First, I'd convert your datatable data into a list of TreeNode objects. Forget about relationships, just create a list with each objects Children being empty. I wrote a method to simulate data retrival from datatable. Instead of that you can use your actual datatable.
static List<DataTableData> GetDataTableData()
{
var data = new List<DataTableData>
{
new DataTableData() { Id = "23456", ParentID = "12345", Name = "Middle" },
new DataTableData() { Id = "55555", ParentID = "12345", Name = "Middle" },
new DataTableData() { Id = "34567", ParentID = "23456", Name = "Bottom" },
new DataTableData() { Id = "12345", ParentID = string.Empty, Name = "Top" },
new DataTableData() { Id = "45678", ParentID = "23456", Name = "Bottom" },
new DataTableData() { Id = "66666", ParentID = "55555", Name = "Bottom" }
};
return data;
}
And this is what your Main() would look like:
static void Main(string[] args)
{
var treeNodes = new List<TreeNode>();
var dataTable = GetDataTableData();
foreach (var data in dataTable)
{
treeNodes.Add(new TreeNode() { Id = data.Id, Name = data.Name, ParentID = data.ParentID });
}
var root = BuildTree(treeNodes);
Console.ReadLine();
}
Now, in my BuildTree() method, instead of passing the datatable I can pass my TreeNode list, and return just the root node.
public static TreeNode BuildTree(List<TreeNode> nodes)
{
foreach (var node in nodes)
{
node.Children = nodes.Where(x => x.ParentID == node.Id).ToList();
}
return nodes.Find(x => x.IsRoot);
}
BuildTree() Breakdown
The nodes list already have all the nodes corresponding to data in your datatable. The BuildTree() is merely going to create the parent-child relations and fill in each object's Children list.
So I iterate through the list, and see what other elements in the list are supposed to be its children. When you have iterated through the list you'd created all the parent-child relationships. Finally, I pick the root node (the one who's ParentId is empty) and return it.
Edit
Here's an easy method to print and verify your tree.
static void PrintTree(TreeNode node, int indents)
{
for (int tab = 0; tab < indents; tab++)
{
Console.Write("\t");
}
Console.WriteLine("{0} - {1}", node.Id, node.Name);
if (node.Children != null && node.Children.Count > 0)
{
indents++;
foreach (var child in node.Children)
{
PrintTree(child, indents);
}
}
}
My output looks like this:

If you are building a class structure then you need a class with a recursive method. Not sure how efficient this will be if it gets too big. Execute the method from the top of the tree.
public class TreeNode
{
public string id { get; set; }
public string name { get; set; }
public string parentID { get; set; }
public List<TreeNode> children { get; set; }
public TreeNode() {
children = new List<TreeNode>();
}
public TreeNode FindParentWithID(string ID)
{
TreeNode ParentWithID = null;
//check my parentID if i am the one being looked for then return
if (id == ID) return this;
//search children
foreach (TreeNode treeNode in children)
{
ParentWithID = treeNode.FindParentWithID(ID);
if (ParentWithID != null)
{
break;
}
}
return ParentWithID;
}
}
You would load your data into the classes from the database. I had to hard code the values for the example to work:
TreeNode treeNode5 = new TreeNode() { id = "ABCD-12345", parentID = null, name = "Top" };
TreeNode treeNode6 = new TreeNode() { id = "ABCD-12346", parentID = "ABCD-12345", name = "Middle" };
treeNode5.children.Add(treeNode6);
TreeNode treeNode7 = new TreeNode() { id = "ABCD-12347", parentID = "ABCD-12346", name = "Bottom" };
TreeNode treeNode8 = new TreeNode() { id = "ABCD-12348", parentID = "ABCD-12346", name = "Bottom" };
treeNode6.children.Add(treeNode7);
treeNode6.children.Add(treeNode8);
TreeNode topOne = treeNode5.FindParentWithID("ABCD-12346");
topOne will be end up being treeNode6 name="Middle" in this example.

try this code, i have same issue and it works perfectly
this method used to build tree from list of items, by looping through all items, and add each item to its parent's child list. and return only the root item with its nested child.
public TreeNode BuildTree(List<TreeNode> source)
{
// build the children list for each item
foreach (var item in source)
{
var itm = source.Where(i => i.parentID == item.Id).ToList();
item.ChildItems = itm;
}
// return only the root parents with its child inside
return source.Where(i => i.parentID == null).FirstOrDefault();
}
noting that this method return only on TreeNode Object with its child, you can return List by changing .FirstOrDefault() to .ToList() in return line

C# linq hierarchy tree

Tag class consists of ID Name and List<Tagging> :
public class Tag
{
public int Id { get; set; }
public string Name { get; set; }
public ICollection<Tagging> Tagging { get; set; }
}
Tagging class :
public class Tagging
{
public int Id { get; set; }
[ForeignKey("ParentTag")]
public int ParentId { get; set; }
public Tag ParentTag { get; set; }
[ForeignKey("ChildTag")]
public int ChildId { get; set; }
public Tag ChildTag { get; set; }
}
Tagging class just express many to many relationship between tags, for hierarchical purpose.
For example given a list :
List<Tag> tags = new List<Tag>();
var parent = new Tag {Name = "Parent", Id = 1, Tagging = new List<Tagging>{ new Tagging{ ParentId = 1, ChildId = 2}}};
var child = new Tag {Name = "Child", Id = 2, Tagging = new List<Tagging> { new Tagging { ParentId = 2, ChildId = 3 }}};
var grandChild = new Tag {Name = "GrandChild", Id = 3};
tags.Add(parent);
tags.Add(child);
tags.Add(grandChild);
I am trying to loop through all hierarchical objects connected to his parent. For example if you call a method getAllHiearchyObject(Tag parent)
Output should be something like this :
Name : "Parent", Id = 1;
Name : "Child", Id : 2;
Name : "GrandChild", Id :3
I need an actual implementation of getAllHiearchyObject(Tag parent)

How about this...
static IEnumerable<Tag> FlattenTag(Tag root)
{
yield return root;
if (root.Tagging != null)
foreach (var childTagging in root.Tagging)
if (childTagging.ChildTag != null)
foreach (var grandChildTag in FlattenTag(childTagging.ChildTag))
yield return grandChildTag;
}
Note that the second foreach above allows for the use of yield with recursion.
Usage...
foreach(var tag in FlattenTag(root))
...

Only one parent to one child.
For a simple case when you have only one parent-child relationship you can create methods like:
public static class EnumerableExtensions
{
#region Methods
public static IEnumerable<T> Unwind<T>(T first, Func<T, T> getNext)
where T : class
{
if (getNext == null)
throw new ArgumentNullException(nameof(getNext));
return Unwind(
first: first,
getNext: getNext,
isAfterLast: item =>
item == null);
}
public static IEnumerable<T> Unwind<T>(
T first,
Func<T, T> getNext,
Func<T, Boolean> isAfterLast)
{
if (getNext == null)
throw new ArgumentNullException(nameof(getNext));
if (isAfterLast == null)
throw new ArgumentNullException(nameof(isAfterLast));
var current = first;
while(!isAfterLast(current))
{
yield return current;
current = getNext(current);
}
}
#endregion
}
And use them in the following way (I have set ChildTag in Taggings, as it will be done by EF):
List<Tag> tags = new List<Tag>();
var grandChild = new Tag { Name = "GrandChild", Id = 3 };
var child = new Tag { Name = "Child", Id = 2, Tagging = new List<Tagging> { new Tagging { ParentId = 2, ChildId = 3, ChildTag = grandChild } } };
var parent = new Tag { Name = "Parent", Id = 1, Tagging = new List<Tagging> { new Tagging { ParentId = 1, ChildId = 2, ChildTag = child } } };
tags.Add(parent);
tags.Add(child);
tags.Add(grandChild);
var fromParent = EnumerableExtensions
.Unwind(
parent,
item =>
item?.Tagging?.FirstOrDefault()?.ChildTag)
.ToArray();
Console.WriteLine("Parent to child:");
foreach (var item in fromParent)
{
Console.WriteLine(item);
}
Proper parent to many children
For a proper tree creation you will have to use:
public class UnwoundItem<T> : IEnumerable<UnwoundItem<T>>
{
private readonly T _item;
private readonly IEnumerable<UnwoundItem<T>> _unwoundItems;
public UnwoundItem(T item, IEnumerable<UnwoundItem<T>> unwoundSubItems)
{
this._item = item;
this._unwoundItems = unwoundSubItems ?? Enumerable.Empty<UnwoundItem<T>>();
}
public T Item
{
get
{
return this._item;
}
}
public IEnumerable<UnwoundItem<T>> UnwoundSubItems
{
get
{
return this._unwoundItems;
}
}
public IEnumerator<UnwoundItem<T>> GetEnumerator()
{
return this._unwoundItems.GetEnumerator();
}
IEnumerator IEnumerable.GetEnumerator()
{
return this.GetEnumerator();
}
}
and
public static class EnumerableExtensions
{
#region Methods
public static UnwoundItem<T> UnwindMany<T>(
T first,
Func<T, IEnumerable<T>> getNext)
where T : class
{
if (getNext == null)
throw new ArgumentNullException(nameof(getNext));
return UnwindMany(
first: first,
getNext: getNext,
isAfterLast: collection =>
collection == null);
}
public static UnwoundItem<T> UnwindMany<T>(
T first,
Func<T, IEnumerable<T>> getNext,
Func<IEnumerable<T>, Boolean> isAfterLast)
{
if (getNext == null)
throw new ArgumentNullException(nameof(getNext));
if (isAfterLast == null)
throw new ArgumentNullException(nameof(isAfterLast));
var currentItems = getNext(first);
if (isAfterLast(currentItems))
return new UnwoundItem<T>(
item: first,
unwoundSubItems: Enumerable.Empty<UnwoundItem<T>>());
return new UnwoundItem<T>(
item: first,
unwoundSubItems: currentItems
.Select(item =>
UnwindMany(
item,
getNext,
isAfterLast)));
}
#endregion
}
It can be tested with:
private static void Print<T>(IEnumerable<UnwoundItem<T>> items, Func<T, String> toString, Int32 level)
{
var indent = new String(' ', level * 4);
foreach (var item in items)
{
Console.Write(indent);
Console.WriteLine(toString(item.Item));
Print(item.UnwoundSubItems, toString, level + 1);
}
}
...
var grandChild = new Tag { Name = "GrandChild", Id = 3 };
var grandChild2 = new Tag { Name = "GrandChild 2", Id = 33 };
var child = new Tag { Name = "Child", Id = 2, Tagging = new List<Tagging> { new Tagging { ParentId = 2, ChildId = 3, ChildTag = grandChild } } };
var child2 = new Tag { Name = "Child 2", Id = 22, Tagging = new List<Tagging> { new Tagging { ParentId = 2, ChildId = 33, ChildTag = grandChild2 } } };
var parent = new Tag { Name = "Parent", Id = 1,
Tagging = new List<Tagging> {
new Tagging { ParentId = 1, ChildId = 2, ChildTag = child },
new Tagging { ParentId = 1, ChildId = 2, ChildTag = child2 } }
};
var fromParent = EnumerableExtensions
.UnwindMany(
parent,
item =>
item?.Tagging?.Select(tagging => tagging.ChildTag));
Console.WriteLine("Parent to child:");
Print(new[] { fromParent }, item => item.Name, 0);

Efficient ordering of hierarchy data with only parentId

Problem
What is the fastest method to order a flat unordered set of nodes such that the parent is always listed before it's children. My current solution uses a queue to iterate the tree in a breadth first manner. However I've been wondering if there was a better / more efficient method.
Notes:
I can't pre compute any values
Id and ParentId could also be Guids (even if ints I can't guarantee sequential ids)
Linq Pad Code
void Main()
{
var nodes = new Node[] {
new Node { Id = 7, ParentId = 3 },
new Node { Id = 4, ParentId = 2 },
new Node { Id = 1, ParentId = 0 },
new Node { Id = 2, ParentId = 1 },
new Node { Id = 3, ParentId = 1 },
new Node { Id = 5, ParentId = 2 },
new Node { Id = 6, ParentId = 3 },
new Node { Id = 8, ParentId = 0 },
};
SortHierarchy(nodes).Dump();
}
IEnumerable<Node> SortHierarchy(IEnumerable<Node> list)
{
// hashtable lookup that allows us to grab references to the parent containers, based on id
var lookup = new Dictionary<int, List<Node>>();
Queue<Node> nodeSet = new Queue<Node>();
List<Node> children;
foreach (Node item in list) {
if (item.ParentId == 0) { // has no parent
nodeSet.Enqueue(item); // This will add all root nodes in the forest
} else {
if (lookup.TryGetValue(item.ParentId, out children)) {
// add to the parent's child list
children.Add(item);
} else {
// no parent added yet
lookup.Add(item.ParentId, new List<Node> { item });
}
}
}
while (nodeSet.Any()) {
var node = nodeSet.Dequeue();
if (lookup.TryGetValue(node.Id, out children)) {
foreach (var child in children) {
nodeSet.Enqueue(child);
}
}
yield return node;
}
}
private class Node {
public int Id { get; set; }
public int ParentId { get; set; }
public string Name { get; set; }
}
Research
I did find this however it wasn't quite what I was after (also the code doesn't work)
Building hierarchy objects from flat list of parent/child

This code gives the same dump() result, order the list with parentid first:
IEnumerable<Node> SortHierarchy2(IEnumerable<Node> list)
{
return list.OrderBy(l => l.ParentId);
}
This will work as long as your child ids are < their parents ids.

How can I construct a tree using this data structure

I have this tree structure, which a node may have multiple nodes.
public class Node
{
public Node()
{
ChildLocations = new HashSet<Node>();
}
public int Id { get; set; }
public string Name { get; set; }
public virtual int? ParentLocationId { get; set; }
public virtual ICollection<Node> ChildLocations { get; set; }
}
Now, I want to add a list of parent-child value into this structure. like:
{1,A} -> {2,B}
{1,A} -> {3,C}
{1,A} -> {4,D}
{3,C} -> {5,E}
{3,C} -> {6,F}
to construct a tree looks like this:
1A
/ | \
2B 3C 4D
/ \
5E 6F
finally, it returns the root reference.
I have come out this solution. But I have no confidence with recursive part. It this correct?
public class Tree
{
Node root;
public Node Root
{
get { return root; }
}
public void Add(int parentId, string parentName, int childId, string childName)
{
if (root == null)
{
root = new Node { Id = parentId, Name = parentName };
root.ChildLocations.Add(new Node { Id = childId, Name = childName });
}
else
{
Add(root, parentId, parentName, childId, childName);
}
}
private void Add(Node node, int parentId, string parentName, int childId, string childName)
{
if (node == null)
{
return;
}
if (node.Id == parentId)
{
node.ChildLocations.Add(new Node { Id = childId, Name = childName });
return;
}
foreach (var n in node.ChildLocations)
{
Add(n, parentId, parentName, childId, childName);
}
}
}

As per my comment to your question, this works to build the tree you require:
public Node BuildTree()
{
var _1A = new Node() { Id = 1, Name = "A", };
var _2B = new Node() { Id = 2, Name = "B", };
var _3C = new Node() { Id = 3, Name = "C", };
var _4D = new Node() { Id = 4, Name = "D", };
var _5E = new Node() { Id = 5, Name = "E", };
var _6F = new Node() { Id = 6, Name = "F", };
_1A.ChildLocations.Add(_2B);
_1A.ChildLocations.Add(_3C);
_1A.ChildLocations.Add(_4D);
_3C.ChildLocations.Add(_5E);
_3C.ChildLocations.Add(_6F);
return _1A;
}
But it's not very general purpose. Can you elaborate on your needs?

C# algorithm for generating hierarchy

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.

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);
var childrenFetcher = default(Func<TKey, IEnumerable<Node<T>>>);
childrenFetcher = parentId => families[parentId]
.OrderBy(orderingKeySelector)
.Select(x => new Node<T>(x, childrenFetcher(keySelector(x))));
return childrenFetcher(topMostKey);
}
Utilizes this small node class:
public class Node<T>
{
public T Value { get; private set; }
public IList<Node<T>> Children { get; private set; }
public Node(T value, IEnumerable<Node<T>> children)
{
this.Value = value;
this.Children = new List<Node<T>>(children);
}
}
It's generic enough to work for a variety of problems, including my text file issue. Nifty!
****UPDATE****: Here's how you'd use it:
// Given some example data:
var items = new[]
{
new Foo()
{
Id = 1,
ParentId = -1, // Indicates no parent
Position = 0
},
new Foo()
{
Id = 2,
ParentId = 1,
Position = 0
},
new Foo()
{
Id = 3,
ParentId = 1,
Position = 1
}
};
// Turn it into a hierarchy!
// We'll get back a list of Node<T> containing the root nodes.
// Each node will have a list of child nodes.
var hierarchy = items.Hierarchize(
-1, // The "root level" key. We're using -1 to indicate root level.
f => f.Id, // The ID property on your object
f => f.ParentId, // The property on your object that points to its parent
f => f.Position, // The property on your object that specifies the order within its parent
);

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

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

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);
}

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

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.

Here is the example that #baran asked for:
var lHierarchicalMenuItems = lMenuItemsFromDB.Hierarchize(0, aItem => aItem.Id, aItem => aItem.ParentId, aItem => aItem.Position);