Using LINQ, how would you filter out all but one item of a particular criteria from a list?

Question

I realize my title probably isn't very clear so here's an example:

I have a list of objects with two properties, A and B.

public class Item
{
    public int A { get; set; }
    public int B { get; set; }
}

var list = new List<Item>
{
    new Item() { A = 0, B = 0 },
    new Item() { A = 0, B = 1 },
    new Item() { A = 1, B = 0 },
    new Item() { A = 2, B = 0 },
    new Item() { A = 2, B = 1 },
    new Item() { A = 2, B = 2 },
    new Item() { A = 3, B = 0 },
    new Item() { A = 3, B = 1 },
}

Using LINQ, what's the most elegant way to collapse all the A = 2 items into the first A = 2 item and return along with all the other items? This would be the expected result.

var list = new List<Item>
{
    new Item() { A = 0, B = 0 },
    new Item() { A = 0, B = 1 },
    new Item() { A = 1, B = 0 },
    new Item() { A = 2, B = 0 },
    new Item() { A = 3, B = 0 },
    new Item() { A = 3, B = 1 },
}

I'm not a LINQ expert and already have a "manual" solution but I really like the expressiveness of LINQ and was curious to see if it could be done better.

Answer 1

How about:

var collapsed = list.GroupBy(i => i.A)
                    .SelectMany(g => g.Key == 2 ? g.Take(1) : g);

The idea is to first group them by A and then select those again (flattening it with .SelectMany ) but in the case of the Key being the one we want to collapse, we just take the first entry with Take(1) .

Answer 2

One way you can accomplish this is with GroupBy . Group the items by A , and use a SelectMany to project each group into a flat list again. In the SelectMany , check if A is 2 and if so Take(1) , otherwise return all results for that group. We're using Take instead of First because the result has to be IEnumerable .

var grouped = list.GroupBy(g => g.A);
var collapsed = grouped.SelectMany(g =>
{
     if (g.Key == 2)
     {
         return g.Take(1);
     }
     return g;
});

Answer 3

One possible solution (if you insist on LINQ):

int a = 2;  
var output = list.GroupBy(o => o.A == a ? a.ToString() : Guid.NewGuid().ToString())
                 .Select(g => g.First())
                 .ToList();

Group all items with A=2 into group with key equal to 2 , but all other items will have unique group key (new guid), so you will have many groups having one item. Then from each group we take first item.

Answer 4

Yet another way:

var newlist = list.Where (l => l.A != 2 ).ToList();
newlist.Add( list.First (l => l.A == 2) );

Answer 5

An alternative to other answers based on GroupBy can be Aggregate :

// Aggregate lets iterate a sequence and accumulate a result (the first arg)
var list2 = list.Aggregate(new List<Item>(), (result, next) => {
     // This will add the item in the source sequence either
     // if A != 2 or, if it's A == 2, it will check that there's no A == 2
     // already in the resulting sequence!
     if(next.A != 2 || !result.Any(item => item.A == 2)) result.Add(next);  
     return result;
});

Answer 6

What about this:

list.RemoveAll(l => l.A == 2 && l != list.FirstOrDefault(i => i.A == 2));

if you whould like more efficient way it would be:

var first = list.FirstOrDefault(i => i.A == 2);
list.RemoveAll(l => l.A == 2 && l != first);