Develop Reference

C# equality with list-based properties

I've read numerous articles related to proper equality in C#:
http://www.loganfranken.com/blog/687/overriding-equals-in-c-part-1/
What is the best algorithm for an overridden System.Object.GetHashCode?
Assume the following sample class:
public class CustomData
{
public string Name { get; set;}
public IList<double> Values = new List<double>();
}
Would it still be the case to compare the Values property using .Equals()? Here is a full equality sample of what I mean:
#region Equality
public override bool Equals(object value)
{
if(Object.ReferenceEquals(null, value)) return false; // Is null?
if (Object.ReferenceEquals(this, value)) return true; // Is the same object?
if (value.GetType() != this.GetType()) return false; // Is the same type?
return IsEqual((CustomData)value);
}
public bool Equals(CustomData obj)
{
if (Object.ReferenceEquals(null, obj)) return false; // Is null?
if (Object.ReferenceEquals(this, obj)) return true; // Is the same object?
return IsEqual(obj);
}
private bool IsEqual(CustomData obj)
{
return obj is CustomData other
&& other.Name.Equals(Name)
&& other.Values.Equals(Values);
}
public override int GetHashCode()
{
unchecked
{
// Choose large primes to avoid hashing collisions
const int HashingBase = (int) 2166136261;
const int HashingMultiplier = 16777619;
int hash = HashingBase;
hash = (hash * HashingMultiplier) ^ (!Object.ReferenceEquals(null, Name) ? Name.GetHashCode() : 0);
hash = (hash * HashingMultiplier) ^ (!Object.ReferenceEquals(null, Values) ? Values.GetHashCode() : 0);
return hash;
}
}
public static bool operator ==(CustomData obj, CustomData other)
{
if (Object.ReferenceEquals(obj, other)) return true;
if (Object.ReferenceEquals(null, obj)) return false; // Ensure that "obj" isn't null
return (obj.Equals(other));
}
public static bool operator !=(CustomData obj, CustomData other) => !(obj == other);
#endregion

List<T>.Equals(List<T> other) will compare references. If you want equality for property Values to be defined as identical sequences of doubles, use the IEnumerable<TSource>.SequenceEqual.(IEnemerable<TSource> other) method (MSDN). See a refactored version of your IsEqual(CustomData obj) below:
private bool IsEqual(CustomData obj)
{
return obj is CustomData other
&& other.Name.Equals(Name)
&& other.Values.SequenceEqual(Values);
}

GetHashCode() override coliding way to often

I'm using unity, and unity does not have a tuple in it, so I created my own tuple class to work since I needed it for my Dictionary.
Dictionary <Tuple<int,int>, Tile>
Tile class that I created and isn't really relevant to solve this problem(at least I think it wont help).
But the problem is that I'm using both negative and positive integer in my tuples, and when I use my current GetHashCode() with the Tuples, sometimes I get the same HashCode, for example Tuple<-10, 8> and Tuple<-9,-10> both gives -172 when I return the hashcode.
Is there any good GetHashCode that wouldn't get me conflicts?
To be honest I'm only using the operator ==, because I need to check if both tuples have the same integers inside of them, if I could get a operator == that only collides when both integer are the same and in the same order, it would solve my problem.
Some other minor problems, I can't get to understand the Equals override, as it is, it is working, but I don't know how well it works, since I kind of changed every single thing until it worked.
public class Tuple<T1, T2>
{
public T1 First { get; private set; }
public T2 Second { get; private set; }
public Tuple(T1 _First, T2 _Second)
{
First = _First;
Second = _Second;
}
public override int GetHashCode()
{
int hash = 0;
hash = First.GetHashCode() * 17 + Second.GetHashCode() + First.GetHashCode();
return hash;
}
public static bool operator==(Tuple<T1, T2> obj1, Tuple<T1, T2> obj2)
{
if (ReferenceEquals(null, obj2))
return false;
return (obj1.GetHashCode() == obj2.GetHashCode());
}
public static bool operator!=(Tuple<T1, T2> obj1, Tuple<T1, T2> obj2)
{
if (ReferenceEquals(null, obj2))
return true;
return !(obj1.GetHashCode() == obj2.GetHashCode());
}
public bool Equals(Tuple<T1, T2> other)
{
if (other == null)
return false;
if (GetHashCode() == other.GetHashCode())
return true;
else
return false;
}
public override bool Equals(object obj)
{
if (ReferenceEquals(null, obj))
{
return false;
}
if (ReferenceEquals(this, obj))
{
return true;
}
Tuple<T1, T2> other = obj as Tuple<T1, T2>;
return obj.GetType() == GetType() && Equals(other);
}
}
public static class Tuple
{
public static Tuple<T1, T2> New<T1, T2>(T1 first, T2 second)
{
var tuple = new Tuple<T1, T2>(first, second);
return tuple;
}
}

GetHashCode() isn't supposed to be collision free. You should use it to determine if two things might be the same objects, and then you have to actually do a thorough check to see if they are.
For example, your == method should be written more like this:
public static bool operator==(Tuple<T1, T2> obj1, Tuple<T1, T2> obj2)
{
if (ReferenceEquals(null, obj2))
return false;
if (obj1.GetHashCode() != obj2.GetHashCode())
{
return false;
}
return DefaultComparer<T1>.Equals(obj1.First, obj2.First) && DefaultComparer<T2>.Equals(obj1.Second, obj2.Second);
}
Also, don't forget to consider the case where obj1 and obj2 are both null.
If you're implementing your own Tuple, you might consider just stealing Microsoft's from the Reference Source repository, or at least use it as a base for your own.

I'm using unity, and unity does not have a tuple in it
It supports Tuple if you have Unity 2017 and above.
Go to Edit --> Project Settings --> Player --> Other Settings --> Configuration --> Scripting Runtime Version --> .NET 4.x Equivalent.
Reload or restart Visual Studio and you should be able to use Tuple. If you are not using Unity 2017 and above and also don't want to update then see John's answer.

This is what resharper generates for you automatically. Just note how they do the GetHashCode() and Equals.
private class Tuple<T1,T2> : IEquatable<Tuple<T1, T2>>
{
public T1 First {get;}
public T2 Second {get;}
public Tuple(T1 first, T2 second)
{
First = first;
Second = second;
}
public bool Equals(Tuple<T1, T2> other)
{
if (ReferenceEquals(null, other)) return false;
if (ReferenceEquals(this, other)) return true;
return EqualityComparer<T1>.Default.Equals(First, other.First) && EqualityComparer<T2>.Default.Equals(Second, other.Second);
}
public override bool Equals(object obj)
{
if (ReferenceEquals(null, obj)) return false;
if (ReferenceEquals(this, obj)) return true;
if (obj.GetType() != this.GetType()) return false;
return Equals((Tuple<T1, T2>) obj);
}
public override int GetHashCode()
{
unchecked
{
return (EqualityComparer<T1>.Default.GetHashCode(First) * 397) ^ EqualityComparer<T2>.Default.GetHashCode(Second);
}
}
public static bool operator ==(Tuple<T1, T2> left, Tuple<T1, T2> right)
{
return Equals(left, right);
}
public static bool operator !=(Tuple<T1, T2> left, Tuple<T1, T2> right)
{
return !Equals(left, right);
}
}

I have found PropertyCompare (*) to be useful in auto-generating Equals implementations. It will automatically compare all public properties (so if you add a new public property you don't need to change anything except GetHashCode (and even that is technically optional).
It uses Cache to be reasonably performant - it takes a one-off hit (per type) to generate the appropriate expressions for the comparisons.
using System;
using System.Linq.Expressions;
namespace YourApp
{
public class Tuple<T1, T2>
{
public T1 First { get; private set; }
public T2 Second { get; private set; }
public Tuple(T1 _First, T2 _Second)
{
First = _First;
Second = _Second;
}
public override int GetHashCode()
{
var hash = 0;
// Implement this however you like
hash = First.GetHashCode() * 17 + Second.GetHashCode() + First.GetHashCode();
return hash;
}
public static bool operator ==(Tuple<T1, T2> x, Tuple<T1, T2> y)
{
return PropertyCompare.Equal(x, y);
}
public static bool operator !=(Tuple<T1, T2> x, Tuple<T1, T2> y)
{
return !PropertyCompare.Equal(x, y);
}
public bool Equals(Tuple<T1, T2> other)
{
return PropertyCompare.Equal(this, other);
}
public override bool Equals(object obj)
{
return PropertyCompare.Equal(this, obj);
}
}
public static class Tuple
{
public static Tuple<T1, T2> New<T1, T2>(T1 first, T2 second)
{
var tuple = new Tuple<T1, T2>(first, second);
return tuple;
}
}
public class Program
{
public static void Main()
{
var bob1 = Tuple.New("a", 1);
var bob2 = Tuple.New("a", 1);
Console.WriteLine(bob1 == bob2);
Console.ReadLine();
}
}
public static class PropertyCompare
{
public static bool Equal<T>(T x, object y) where T : class
{
return Cache<T>.Compare(x, y as T);
}
public static bool Equal<T>(T x, T y)
{
if (x == null)
{
return y == null;
}
if (y == null)
{
return false;
}
return Cache<T>.Compare(x, y);
}
private static class Cache<T>
{
internal static readonly Func<T, T, bool> Compare;
static Cache()
{
var props = typeof(T).GetProperties();
if (props.Length == 0)
{
Compare = delegate { return true; };
return;
}
var x = Expression.Parameter(typeof(T), "x");
var y = Expression.Parameter(typeof(T), "y");
Expression body = null;
for (var i = 0; i < props.Length; i++)
{
var propEqual = Expression.Equal(
Expression.Property(x, props[i]),
Expression.Property(y, props[i]));
if (body == null)
{
body = propEqual;
}
else
{
body = Expression.AndAlso(body, propEqual);
}
}
Compare = Expression.Lambda<Func<T, T, bool>>(body, x, y)
.Compile();
}
}
}
}
(*) I found it online somewhere, alas I can't remember where and Google is failing me here.

Checking against null in Equals()

How come if I comment out the second line when overriding Equals() like so:
public override bool Equals(object obj) {
if(object.ReferenceEquals(this, obj)) return true;
//if(obj == null) return false;
Person other = obj as Person;
if(other == null) return false;
return this.Name == other.Name;
}
I get a NullReferenceException? If I uncomment it, it'll work. Also I made sure that the obj argument wasn't a null, and it still does that.
Here is full code
namespace MyNameSpace{
class Person : IComparable<Person>{
public string Name { get; set; }
public Person(string name) {
Name = name;
}
public static bool operator <(Person x, Person y) {
return x.CompareTo(y) < 0;
}
public static bool operator >(Person x, Person y) {
return x.CompareTo(y) > 0;
}
public static bool operator ==(Person x, Person y) {
return x.Equals(y);
}
public static bool operator !=(Person x, Person y) {
return !x.Equals(y);
}
public override bool Equals(object obj) {
if(object.ReferenceEquals(this, obj)) return true;
//if(obj == null) return false;
Person other = obj as Person;
if(other == null) return false;
return this.Name == other.Name;
}
public int CompareTo(Person other) {
return this.Name.CompareTo(other.Name);
}
}
class Program {
static void Main(string[] args) {
Person one = new Person("one");
Person two = new Person("two");
Console.WriteLine(one == two);
}
}
}

I suspect you have a custom == operator on Person, which is being invoked by the line:
if(other == null) return false;
Which also hints that the operator is buggy and should be fixed.
Edit: and with your update: here is the buggy custom == operator:
public static bool operator ==(Person x, Person y) {
return x.Equals(y);
}
Using just:
public static bool operator ==(Person x, Person y) {
return Equals(x,y);
}
would solve that, along with:
public override bool Equals(object obj) {
if(obj == (object)this) return true; // ref equality, the cheap way
if(obj is Person) {
Person other = (Person)obj;
return this.Name == other.Name;
}
return false;
}

As a general rule for IComparable implementation, I would strongly recommend Eric Lippert's approach. It's very simple and helps a lot to not get NRE.
Basically, you don't call either Equals or == in any operator overload. You just call a unique static method that does all the job:
public int CompareTo(Natural x) { return CompareTo(this, x); }
public static bool operator <(Natural x, Natural y) { return CompareTo(x, y) < 0; }
public static bool operator >(Natural x, Natural y) { return CompareTo(x, y) > 0; }
public static bool operator <=(Natural x, Natural y) { return CompareTo(x, y) <= 0; }
public static bool operator >=(Natural x, Natural y) { return CompareTo(x, y) >= 0; }
public static bool operator ==(Natural x, Natural y) { return CompareTo(x, y) == 0; }
public static bool operator !=(Natural x, Natural y) { return CompareTo(x, y) != 0; }
public override bool Equals(object obj) { return CompareTo(this, obj as Natural) == 0; }
public bool Equals(Natural x) { return CompareTo(this, x) == 0; }
// negative means x < y
// positive means x > y
// zero means x == y
// two nulls are equal
// otherwise, null is always smaller
private static int CompareTo(Natural x, Natural y) {
if (ReferenceEquals(x, y))
return 0;
else if (ReferenceEquals(x, null))
return -1;
else if (ReferenceEquals(y, null))
return 1;
else if (ReferenceEquals(x, Zero))
return -1;
else if (ReferenceEquals(y, Zero))
return 1;
else if (x.head == y.head)
return CompareTo(x.tail, y.tail);
else if (x.head == ZeroBit)
return CompareTo(x.tail, y.tail) > 0 ? 1 : -1;
else
return CompareTo(x.tail, y.tail) < 0 ? -1 : 1;
}

I get it, I was recursively calling operator ==, and at some point i had operator == (x, y), where x and y were both null, and then I did x.Equals(), that's why it threw that exception.

Distinct values from List of objects

I need your help. I am trying to get distinct values from List of objects.
My class looks like this:
class Chromosome
{
public bool[][] body { get; set; }
public double fitness { get; set; }
}
Now I have List<Chromosome> population. And now what I need is a way, how I can get new list: List<Chromosome> newGeneration. This new list will contain only unique chromosomes from original list - population.Chromosome is unique, when his whole body (which in this case is 2D bool array) is unique in comparison to the other chromosomes.
I know, that there is something like MoreLINQ, but I am not sure, whether I should use 3rd party code and I know that I should overwrite some methods, but I am kind of lost. So I would really appreciate some nice step by step description, that even idiot could accomplish :)
THX

First, implement the equality operator (this goes into class Chromosome):
public class Chromosome : IEquatable<Chromosome>
{
public bool[][] body { get; set; }
public double fitness { get; set; }
bool IEquatable<Chromosome>.Equals(Chromosome other)
{
// Compare fitness
if(fitness != other.fitness) return false;
// Make sure we don't get IndexOutOfBounds on one of them
if(body.Length != other.body.Length) return false;
for(var x = 0; x < body.Length; x++)
{
// IndexOutOfBounds on inner arrays
if(body[x].Length != other.body[x].Length) return false;
for(var y = 0; y < body[x].Length; y++)
// Compare bodies
if(body[x][y] != other.body[x][y]) return false;
}
// No difference found
return true;
}
// ReSharper's suggestion for equality members
public override bool Equals(object obj)
{
if (ReferenceEquals(null, obj))
{
return false;
}
if (ReferenceEquals(this, obj))
{
return true;
}
if (obj.GetType() != this.GetType())
{
return false;
}
return this.Equals((Chromosome)obj);
}
public override int GetHashCode()
{
unchecked
{
return ((this.body != null ? this.body.GetHashCode() : 0) * 397) ^ this.fitness.GetHashCode();
}
}
}
Then, use Distinct:
var newGeneration = population.Distinct().ToList();

public class ChromosomeBodyComparer : IEqualityComparer<Chromosome>
{
private bool EqualValues(bool[][] left, bool[][] right)
{
if (left.Length != right.Length)
{
return false;
}
return left.Zip(right, (x, y) => x.SequenceEquals(y)).All();
}
public bool Equals(Chromosome left, Chromosome right)
{
return EqualValues(left.body, right.body)
}
//implementing GetHashCode is hard.
// here is a rubbish implementation.
public int GetHashCode(Chromosome c)
{
int numberOfBools = c.body.SelectMany(x => x).Count();
int numberOfTrues = c.body.SelectMany(x => x).Where(b => b).Count();
return (17 * numberOfBools) + (23 * numberOfTrues);
}
}
Called by:
List<Chromosome> nextGeneration = population
.Distinct(new ChromosomeBodyComparer())
.ToList();

c# similarity help

Anyone can explain these codes for me?
public class SimilarityImages : IComparer<SimilarityImages>, IComparable
{
private readonly ComparableImage source;
private readonly ComparableImage destination;
private readonly double similarity;
public SimilarityImages(ComparableImage source, ComparableImage destination, double similarity)
{
this.source = source;
this.destination = destination;
this.similarity = similarity;
}
public ComparableImage Source
{
get
{
return source;
}
}
public ComparableImage Destination
{
get
{
return destination;
}
}
public double Similarity
{
get
{
return Math.Round(similarity * 100, 1);
}
}
public static int operator !=(SimilarityImages value, SimilarityImages compare)
{
return value.CompareTo(compare);
}
public static int operator <(SimilarityImages value, SimilarityImages compare)
{
return value.CompareTo(compare);
}
public static int operator ==(SimilarityImages value, SimilarityImages compare)
{
return value.CompareTo(compare);
}
public static int operator >(SimilarityImages value, SimilarityImages compare)
{
return value.CompareTo(compare);
}
public override string ToString()
{
return string.Format("{0}, {1} --> {2}", source.File.Name, destination.File.Name, similarity);
}
#region IComparer<SimilarityImages> Members
public int Compare(SimilarityImages x, SimilarityImages y)
{
return x.similarity.CompareTo(y.similarity);
}
#endregion
#region IComparable Members
public int CompareTo(object obj)
{
SimilarityImages other = (SimilarityImages)obj;
return this.Compare(this, other);
}
#endregion
public override bool Equals(object obj)
{
if (obj == null || GetType() != obj.GetType())
{
return false;
}
var other = (SimilarityImages)obj;
var equals = Source.File.FullName.Equals(other.Source.File.FullName, StringComparison.InvariantCultureIgnoreCase);
if (!equals)
{
return false;
}
equals = Destination.File.FullName.Equals(other.Destination.File.FullName, StringComparison.InvariantCultureIgnoreCase);
if (!equals)
{
return false;
}
return true;
}
public override int GetHashCode()
{
return string.Format("{0};{1}", Source.File.FullName, Destination.File.FullName).GetHashCode();
}
}

It looks kind of like String.Compare()
(Comparison of strings by value.)
however the return of integers for the operators <, >, <=, >=, ==, != is a bad practice - user expect a boolean return value.
Also, if the "images" are what I think they are (images as in a matrix of pixels), then returning a value for comparison is not intuitive - What comparison algorithm are you using? - Why are you providing only one comparison algorithm and hiding its name? - What if the images are not of same size or pixel format?