C# LINQ Distinct(f=>f.propertyname) doesnt work

Question

I am trying to write a LINQ query on a collection of a complex type. I want to write a distinct query on this collection for a combination of two fields.

I don't think Lambda expression support piping of Distinct(f=>f.propertyname). Wish it did. Anyone used a simpler implementation then using a Comparer?

It's included in [MoreLINQ](https://www.nuget.org/packages/morelinq/1.0.16006) as [`DistinctBy`, here's the sample source](https://code.google.com/p/morelinq/source/browse/MoreLinq/DistinctBy.cs) — Patryk Ćwiek, Mar 13 '14 at 08:37

Ufuk Hacıoğulları · Accepted Answer · 2014-03-13T09:01:03.737

5

You can use DistinctBy method in MoreLINQ package.

var result = items.DistinctBy(f => f.PropertyName);

You can DistinctBy with anonymous types to get distinct results by two columns.

var result = items.DistinctBy(f => new { f.Property1, f.Property2});

edited Mar 13 '14 at 09:01

answered Mar 13 '14 at 08:39

Ufuk Hacıoğulları

37,978
12
114
156

1

`DISTINCTBY` is not standard LINQ feature – Amit Mar 13 '14 at 08:40
1

@AmitAgrawal who says it should be standard feature? – Sergey Berezovskiy Mar 13 '14 at 08:41
2

@AmitAgrawal I already said it's in MoreLINQ package. – Ufuk Hacıoğulları Mar 13 '14 at 08:42
You can not call directly `DISTINCTBY`. So please don't suggest as such answer until OP is not asking... – Amit Mar 13 '14 at 08:42
2

@AmitAgrawal Well he didn't state that he can't install a NuGet package. It's easier than writing a custom equality comparer. – Ufuk Hacıoğulları Mar 13 '14 at 08:43
1

@AmitAgrawal Let us suggest options, its op's call to use it or not. Who knows it may be useful for someone else in future. – Sriram Sakthivel Mar 13 '14 at 08:54

Sergey Berezovskiy · Answer 2 · 2014-03-13T09:00:01.677

With pure LINQ you can group by all properties you need and then select first item from each group:

var result = items.GroupBy(i => new { i.Prop1, i.Prop2 })
                  .Select(g => g.First());

Pros:

You don't need to create custom comparer, or modify your class, or reference third party library.
It can be translated to SQL and executed on server side if you are querying items from database

Cons:

With Linq to Objects MoreLINQ approach is more efficient (though you should not do premature optimizations)
MoreLINQ syntax is more expressive

score 0 · Answer 3 · answered Mar 13 '14 at 08:37

0

You cannot use distinct like that. Compiler couldn't know which of the multiple values to take. You can use GroupBy() to combine them or use Min(), Max(), First(), Last() or similar to select one from the list depending on your need.

answered Mar 13 '14 at 08:37

Fedor Hajdu

4,657
3
32
50

Tim Schmelter · Answer 4 · 2014-03-13T09:18:26.427

You have several options:

You could either implement a custom IEqualityComparer<T> for the overload of Distinct
or override Equals + GethashCode in your class

Another (less efficient) approach which doesn't need to create a new class or modify an existing is to use the builtin GetHashCode+Equals of an anonymous type and Enumerable.GroupBy:

IEnumerable<Complex> distinctObjects = 
    from c in collection
    group c by  new { c.Prop1, c.Prop2 } into g 
    select g.First(); // change logic if you don't want an arbitrary object(first)

Here's an example of the second approach:

public class Complex
{
    public string Prop1 { get; set; }
    public int Prop2 { get; set; }
    public Complex Prop3 { get; set; }

    public override bool Equals(object obj)
    {
        Complex c2 = obj as Complex;
        if (obj == null) return false;
        return Prop1 == c2.Prop1 && Prop2 == c2.Prop2;
    }

    public override int GetHashCode()
    {
        unchecked // Overflow is fine, just wrap
        {
            int hash = 17;
            hash = hash * 23 + Prop1.GetHashCode();
            hash = hash * 23 + Prop2;
            return hash;
        }
    }
}

The Equals and GethashCode of the IEqualityComparer-class (1. approach) would be similar.

score -2 · Answer 5 · answered Mar 13 '14 at 08:53

You could create an anonymous type on the fly with your two properties, and then use distinct.

Like this:

[Test]
public void Distinct_with_anonymous_type()
{
    var items = new[]
    {
        new {p1 = 'a', p2 = 2, p3=10},   // double over p1 and p2
        new {p1 = 'a', p2 = 3, p3=11}, 
        new {p1 = 'a', p2 = 2, p3=12},   // double over p1 and p2
    };

    var expected = new[]
    {
        new {p1 = 'a', p2 = 2},
        new {p1 = 'a', p2 = 3}, 
    };

    var distinct = items.Select(itm => new { itm.p1, itm.p2 }).Distinct();

    Assert.That(distinct, Is.EquivalentTo(expected));
}

Hope this helps!

C# LINQ Distinct(f=>f.propertyname) doesnt work

5 Answers5