Sorting keys and mapreduce in couchdb

I am doing a scaled down view for counting user comments for a date range and I am facing a problem not solving the problem. The date range selection is ok - I create an array key [YYYY, MM, DD, user_id].

Sample document:

{
   "_id": "1",
   "_rev": "1-84d3fa5f259c7b30c74028ca60a45f91",
   "body": "note text",
   "user_id": 666,
   "created_at": "Thu Apr 26 13:50:20 +0200 2010",
}

      

        
        
        
      

    

Map function:

function(doc){
   var created = new Date(doc.created_at);
   emit([created.getFullYear(),created.getMonth()+1,created.getDate(),doc.user_id],doc);
}

      

        
        
        
      

    

Decrease function:

_count

      

        
        
        
      

    

So far this is OK - when I want to select all the notes for a selected date range, for example from 2010-02-02 to 2010-03-03:

reduce_view_name?startkey=[2010,02,01]&endkey=[2010,02,28,{}]&group_level=0

      

        
        
        
      

    

Easy to select notes / user / day sums - I change group_level to 4. But what about if I want to count notes for a selected date range with one note sum per user? In this case, I get 28 rows per user, but what about 300 users? It will be 300x28 = 7400 lines (if everyone writes every day :).

The second option replaces the user_id from the last place to the first, but in this case, I need to know the id_user and make a selection for each separately.

Does anyone know how to solve it? :)