Using Linq to group and sum with multiple columns

[tshad]

I am trying to figure out how to take some data from a datatable or class that could have 5 columns or could have 10 or 20 columns. This was created pivoting a table which gave me an unknown number of columns. But I can't figure out how to use Linq to handle this. Here is a sample set of code I have been playing with.

         var books = new List<Book>();

         // Add test data
         books.Add(new Book { Author = "Douglas Adams", Title = "The Hitchhiker's Guide to the Galaxy", Genre = "Fiction", Price = 159.95M });
         books.Add(new Book { Author = "Scott Adams", Title = "The Dilbert Principle", Genre = "Fiction", Price = 23.95M });
         books.Add(new Book { Author = "Douglas Coupland", Title = "Generation X", Genre = "Fiction", Price = 300.00M });
         books.Add(new Book { Author = "Walter Isaacson", Title = "Steve Jobs", Genre = "Biography", Price = 219.25M });
         books.Add(new Book { Author = "Michael Freeman", Title = "The Photographer's Eye", Genre = "Photography", Price = 195.50M });

         // Group the books by Genre
         var booksGrouped = from b in books
                            group b by b.Genre into g
                            select new Group<string, Book> { Key = g.Key, Values = g };

         List<Group<string, Book>> bookList = booksGrouped.ToList();

         // Group the books by Genre count number of books
         var booksGroupedCounted = from b in books
                                   group b by b.Genre into g
                                   select new GroupCount<string, Book, int> { Key = g.Key, Values = g, Count = g.Count() };

         List<GroupCount<string, Book, int>> booksCounted = booksGroupedCounted.ToList();

         //Group the books by Genre count number of books and sum the prices
         var booksGroupedCountedSummed = from b in books
                                         group b by b.Genre into g
                                         select new GroupCountSum<string, Book, int, decimal> { Key = g.Key, Values = g, Count = g.Count(), Summed = g.Sum(x => x.Price) };

         List<GroupCountSum<string, Book, int, decimal>> booksCountedSummed = booksGroupedCountedSummed.ToList();

The list<> objects are set up to handle the different number of columns:

   public class Group<K, T>
   {
      public K Key;
      public IEnumerable<T> Values;
   }

   public class GroupCount<K, T, C>
   {
      public K Key;
      public IEnumerable<T> Values;
      public int Count;
   }

   public class GroupCountSum<K, T, C, D>
   {
      public K Key;
      public IEnumerable<T> Values;
      public int Count;
      public decimal Summed;
   }

The IEnumberable<> values is the grouped detail that is put into grouped item. The count and sum fields are the ones that are dynamic. But this can get unwieldy. I would need one for each possibility and have a different linq statement for each one.

I thought about have a list<> object something like:

      public class GroupCountSum<K, T, C1, C2, C3, C4, C5, C6, C7, C8, C9, C10, C11, C12, C13, C14, C15, C16, C17, C18, C19, C20, C21>
      {
         public K Key;
         public IEnumerable<T> Values;
         public int Count1;
         public int Count2;
         public int Count3;
         public int Count4;
         public int Count5;
           ...
      }

But couldn't think about how I would do the select statement since most of the time there would be only 7 or 8.

Any way to change the Linq query to handle this?

Thanks,

Tom