Selecting Items with the same subcollection using LINQ

Question

Here is the pseudo case:

class parent{
   string name; //Some Property
   List<int> myValues;

   .......
}
........
//Initialize some parent classes

List<parent> parentList = new List<parent>();
parentList.add(parent123); //parent123.myValues == {1,2,3}
parentList.add(parent456); //parent456.myValues == {4,5,6}
parentList.add(parentMatch); //parentMatch.myValues == {1,2,3}

What I am aiming for is a query which retrieves a List of parent objects where their myValues Lists are equivalent. In this case it would return parent123 and parentMatch.

Answer 1

Very silly solution:

var groups = list.GroupBy(p => string.Join(",", p.list.Select(i => i.ToString()).ToArray()))
                    .Where(x => x.Count() > 1).ToList();

Result:

an IEnumerable of groups containing parent objects having list with same int (in the same order).

If you need to match list of elements in any order (i.e. 1,2,3 == 3,1,2), just change p.list to p.list.OrderBy(x => x).

Plus, if you're targeting framework 4.0, you can avoid ToArray and ToString

---

EDIT:

added a where to filter single-occurrence groups.

Now if you have these parents:

parent  A  1,2,3
parent  B  1,2,3
parent  C  1,2,3
parent  D  4,5,6
parent  E  4,5,6
parent  F  7,8,9

it returns:

(A,B,C) - (D,E)

Answer 2

Although the "best" solution has already been found, this is my approach: The XOR ^ Operator (C# Reference) is used to get a hash code for the values sequence, thereby the order does not matter and you do not need a parents-to-parents cross join, no Distinct etc.. You can replace the SelectMany with whatever you want to do with the IGrouping. This is only fail safe for distinct values, because XOR with the same hash code will neutralize them.

var dublicates = parentList 
    .GroupBy(p => p.myValues
        .Select(x => x.ToString().GetHashCode())
        .Aggregate((v1, v2) => v1 ^ v2))
    .Where(g => g.Take(2).Count() == 2)
    .SelectMany(g => g);

Answer 3

Try this:

var matches = (from p1 in parentList
               from p2 in parentList
               let c1 = p1.myValues
               let c2 = p2.myValues
               where p1 != p2 &&
                     c1.All(child => c2.Contains(child)) &&
                     c2.All(child => c1.Contains(child))
               select p1).Distinct();

Answer 4

So you can wrap the logic up and just use GroupBy if you implement an IEqualityComparer:

class IntegerListComparer : IEqualityComparer<List<int>>
{
    #region IEqualityComparer<List<int>> Members

    public bool Equals(List<int> x, List<int> y)
    {
        //bool xContainsY = y.All(i => x.Contains(i));
        //bool yContainsX = x.All(i => y.Contains(i));
        //return xContainsY && yContainsX;
        return x.SequenceEqual(y);
    }

    public int GetHashCode(List<int> obj)
    {
        return 0;
    }

    #endregion
}

Call it like so:

var results = list
    .GroupBy(p => p.MyValues, new IntegerListComparer())
    .Where(g => g.Count() > 1)
    .SelectMany(g => g);