Grouping Contiguous Dates

Question

I have a List dates;

I have a class that has:

class NonWorkingDay
{
   public DateTime Start;
   public int Days;
}
         


        

Answer 1

So, we'll start out with this generic iterator function. It takes a sequence and a predicate that accepts two items and returns a boolean. It will read in items from the source and while an item, along with it's previous item, returns true based on the predicate, that next item will be in the "next group". If it returns false, the previous group is full and the next group is started.

public static IEnumerable> GroupWhile(this IEnumerable source
    , Func predicate)
{
    using (var iterator = source.GetEnumerator())
    {
        if (!iterator.MoveNext())
            yield break;

        List currentGroup = new List() { iterator.Current };
        while (iterator.MoveNext())
        {
            if (predicate(currentGroup.Last(), iterator.Current))
                currentGroup.Add(iterator.Current);
            else
            {
                yield return currentGroup;
                currentGroup = new List() { iterator.Current };
            }
        }
        yield return currentGroup;
    }
}

We'll also need this simple helper method that gets the next working day based on a date. If you want to incorporate holidays as well it goes from trivial to quite hard, but that's where the logic would go.

public static DateTime GetNextWorkDay(DateTime date)
{
    DateTime next = date.AddDays(1);
    if (next.DayOfWeek == DayOfWeek.Saturday)
        return next.AddDays(2);
    else if (next.DayOfWeek == DayOfWeek.Sunday)
        return next.AddDays(1);
    else
        return next;
}

Now to put it all together. First we order the days. (If you ensure they always come in ordered you can remove that part.) Then we group the consecutive items while each item is the next work day of the previous.

Then all we need to do is turn an IEnumerable of consecutive dates into a NonWorkingDay. For that the start date is the first date, and Days is the count of the sequence. While normally using both First and Count would iterate the source sequence twice, we happen to know that the sequence returned by GroupWhile is actually a List under the hood, so iterating it multiple times is not a problem, and getting the Count is even O(1).

public IEnumerable GetContiguousDates(IEnumerable dates)
{
    return dates.OrderBy(d => d)
            .GroupWhile((previous, next) => GetNextWorkDay(previous).Date == next.Date)
            .Select(group => new NonWorkingDay
                {
                    Start = group.First(),
                    Days = group.Count(),
                });
}