Skip to content

Add Huffman in C#. #82

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 8 commits into from
May 29, 2018
Merged
Show file tree
Hide file tree
Changes from 1 commit
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
185 changes: 185 additions & 0 deletions chapters/data_compression/huffman/code/cs/HuffmanCoding.cs
Original file line number Diff line number Diff line change
@@ -0,0 +1,185 @@
// submitted by Julian Schacher (jspp)
using System;
using System.Collections.Generic;
using System.Linq;

namespace HuffmanCoding
{
public class EncodeResult
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Maybe a better name would be EncodingResult

{
public List<bool> BitString { get; set; }
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This isn't much more efficient than a regular String and it's much more annoying to debug and print. If you seriously want to produce a packed binary result then I'd go with System.Collections.BitArray but I think a regular String works better for educational purposes.

Copy link
Member Author

@june128 june128 Apr 22, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I will go for the BitArray then. I don't think a string is fitting in this case, since no real compression would be achieved.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The point isn't compression, it's showcasing thr algorithm. And there was no compression achieved with List<bool> either. I'd just go with String (as almost all other implementations in the AAA do).

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You mean since we usually save whole bytes anyway?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Strings in C# are UTF16 encoded which is 2 bytes. If you save the bitstring as a List<bool> that's N bytes, where N is the length of the bitstring. That means you literally only have any compression if the bitstring is a single bit long.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I still think just having this be a string would make things a lot easier

public Dictionary<char, List<bool>> Dictionary { get; set; }
public HuffmanCoding.Node Tree { get; set; }

public EncodeResult(List<bool> bitString, Dictionary<char, List<bool>> dictionary, HuffmanCoding.Node tree)
{
this.BitString = bitString;
this.Dictionary = dictionary;
this.Tree = tree;
}
}

public static class HuffmanCoding
{
// The Node class used for the Huffman Tree.
public class Node
{
public Node[] Children { get; set; } = new Node[2];
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Just have two children, Left and Right. This only makes it easier to misuse the class

public List<bool> BitString { get; set; } = new List<bool>();
public int Weight { get; set; }
public string Key { get; set; }
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

char would work here

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

For the key?

Copy link
Contributor

@zsparal zsparal Apr 22, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes. Maybe char? so you can set it to null in the branches

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The key of a node is string, so that a parent node's/branch's key is a combination of all it's children's keys.


public Node(string key, int weight)
{
this.Key = key;
this.Weight = weight;
}
}

// Node with biggest value at the top.
class NodePriorityList
{
public List<Node> Nodes { get; private set; } = new List<Node>();

public NodePriorityList() { }
public NodePriorityList(List<Node> nodes) => Nodes = nodes.OrderByDescending(n => n.Weight).ToList();
Copy link
Contributor

@zsparal zsparal Apr 21, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Have Node implement IComparable and just do:

public NodePriorityList(List<Node> nodes)
{
    Nodes = nodes.ToList();
    Nodes.Sort();
}


public void AddNode(Node newNode)
{
if (Nodes.Count == 0)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Just have Node implement IComparable<Node> then use something like:

var insertAt = Math.Max(Nodes.BinarySearch(newNode), 0);
Nodes.Insert(insertAt, newNode)

{
Nodes.Add(newNode);
return;
}
for (int i = Nodes.Count - 1; i >= 0; i--)
{
if (Nodes[i].Weight > newNode.Weight)
{
Nodes.Insert(i + 1, newNode);
return;
}
else if (i == 0)
Nodes.Insert(0, newNode);
}
}
}

public static EncodeResult Encode(string input)
{
var root = CreateTree(input);
var dictionary = CreateDictionary(root);
var bitString = CreateBitString(input, dictionary);

return new EncodeResult(bitString, dictionary, root);
}

public static string Decode(EncodeResult result)
{
var output = "";
Node currentNode = result.Tree;
foreach (var boolean in result.BitString)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You should probably name this bit

{
// Go down the tree.
if (!boolean)
currentNode = currentNode.Children[0];
else
currentNode = currentNode.Children[1];

// Check if it's a leaf node.
if (currentNode.Key.Count() == 1)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Just check if all of the children are null. That's how you determine the leaves in a binary tree

Copy link
Contributor

@zsparal zsparal May 28, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You should really not check for leaf nodes this way. It works when the alphabet of whatever you want to encode consists of single characters. You can also run Huffman coding on words, in which case this would faily miserably. The nice solution is to add something like this to Node:

public IsLeaf => LeftChild == null && RightChild == null;

// Then you can just do
if (currentNode.IsLeaf)
{
    // ...
}

{
output += currentNode.Key;
currentNode = result.Tree;
}
}
return output;
}

private static Node CreateTree(string input)
{
// Create a List of all characters and their count in input by putting them into nodes.
var nodes = new List<Node>();
foreach (var character in input)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There's a nicer way to do this. Pseudo-ish code with a toy example:

            var input = "Hello, World";
            var nodes = input
                .GroupBy(c => c)
                .Select(n => new Node { Key = n.Key, Weight = n.Count() })
                .ToList();

{
var result = nodes.Where(n => n.Key[0] == character).SingleOrDefault();

if (result == null)
nodes.Add(new Node(character.ToString(), 1));
else
result.Weight++;
}
// Convert list of nodes to a NodePriorityList.
var nodePriorityList = new NodePriorityList(nodes);
nodes = nodePriorityList.Nodes;
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Don't do this, just use the queue. In fact, just make Nodes private

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

So the Node class?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No, the Nodes property on NodePriorityList


// Create Tree.
while (nodes.Count > 1)
{
var parentNode = new Node("", 0);
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You should have a default constructor so you don't need to pass dummy parameters. I would actually have something like:

public class Node
{
    public static Node CreateLeaf(char key, int weight) { return new Node(...); }
    public static Node CreateBranch(Node left, Node right) { return new Node(...); }
    private Node(...) { /* regular Node constructor with every parameter */ }
}

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I could also create two constructors for the Node class. One that takes key and weight and another one that takes a leftChild and rightChild. Your approach seems more declarative tho. What do you think?

// Add the two nodes with the smallest weight to the parent node and remove them from the tree.
for (int i = 0; i < 2; i++)
{
parentNode.Children[i] = nodes.Last();
Copy link
Contributor

@zsparal zsparal Apr 21, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Make a nice method for this on the priority queue (and don't use the regular list at this point). Something like:

var left = nodeQueue.Pop();
var right = nodeQueue.Pop();
nodeQueue.AddNode(Node.CreateBranch(left, right)); // the weight will be calculated in `CreateBranch` as `left.Weight + right.Weight`

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What are the improvements with using a queue? If I use it BinarySearch() wouldn't be available. I mean, I can re-implement that probably, but if the queue doesn't provide improvements, I see no reason to do so.

parentNode.Key += nodes.Last().Key;
parentNode.Weight += nodes.Last().Weight;

nodes.RemoveAt(nodes.Count - 1);
};
nodePriorityList.AddNode(parentNode);
if (parentNode.Weight > 100)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This shouldn't be here, the weights are not percentages

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

That's left from debugging and I forgot to remove it :S

throw new Exception();
}

return nodePriorityList.Nodes[0];
}

private static Dictionary<char, List<bool>> CreateDictionary(Node root)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I know I'm usually a big proponent for performance, but seriously, just use the recursive method here. With C# 7's local functions you can even make a nice, local function for the recursive part

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hmm, I'll take a look, but I found the non-recursive version easy to understand. Didn't tried the recursive one tho.

{
var dictionary = new Dictionary<char, List<bool>>();

var stack = new Stack<Node>();
stack.Push(root);
Node temp;

while (stack.Count != 0)
{
temp = stack.Pop();

if (temp.Key.Count() == 1)
dictionary.Add(temp.Key[0], temp.BitString);
else
{
for (int i = 0; i < temp.Children.Count(); i++)
{
if (temp.Children[i] != null)
{
if (i == 0)
{
temp.Children[i].BitString.AddRange(temp.BitString);
temp.Children[i].BitString.Add(false);
}
else
{
temp.Children[i].BitString.AddRange(temp.BitString);
temp.Children[i].BitString.Add(true);
}

stack.Push(temp.Children[i]);
}
}
}
}

return dictionary;
}

private static List<bool> CreateBitString(string input, Dictionary<char, List<bool>> dictionary)
{
var bitString = new List<bool>();
foreach (var character in input)
bitString.AddRange(dictionary[character]);

return bitString;
}
}
}
40 changes: 40 additions & 0 deletions chapters/data_compression/huffman/code/cs/Program.cs
Original file line number Diff line number Diff line change
@@ -0,0 +1,40 @@
// submitted by Julian Schacher (jspp)
using System.Collections;
using System.Collections.Generic;

namespace HuffmanCoding
{
class Program
{
static void Main(string[] args)
{
var result = HuffmanCoding.Encode("aaaabbbccd");
// Print dictionary.
foreach (var entry in result.Dictionary)
{
var bitString = "";
Copy link
Contributor

@zsparal zsparal May 28, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is just: string.Join("", entry.Value.Select(bit => bit ? '1' : '0')), but if you take the advice of just using strings, this whole thing is unnecessary.

foreach (var value in entry.Value)
{
if (value)
bitString += "1";
else
bitString += "0";
}
System.Console.WriteLine(entry.Key + " " + bitString);
}
// Print bitString.
var readableBitString = "";
foreach (var boolean in result.BitString)
{
if (boolean)
readableBitString += "1";
else
readableBitString += "0";
}
System.Console.WriteLine(readableBitString);

var originalString = HuffmanCoding.Decode(result);
System.Console.WriteLine(originalString);
}
}
}
6 changes: 6 additions & 0 deletions chapters/data_compression/huffman/huffman.md
Original file line number Diff line number Diff line change
Expand Up @@ -86,4 +86,10 @@ Whether you use a stack or straight-up recursion also depends on the language, b
{% sample lang="hs" %}
### Haskell
[import, lang:"haskell"](code/haskell/huffman.hs)
{% sample lang="cs" %}
### C# #
HuffmanCoding.cs
[import, lang:"csharp"](code/cs/HuffmanCoding.cs)
Program.cs
[import, lang:"csharp"](code/cs/Program.cs)
{% endmethod %}