Efficient week statistics for a QuerySet

Question

I am working on an open source Django time tracking app, Djime, and I'm trying to come up with a more efficient way to produce statistics. So far, we've had some rather long-winded procedural code that would get all TimeSlices for a period, and collate them together in a huge nested list/dictionary mess.

What I'd like to do is to set up a more efficient system – an object or function that would take a QuerySet of TimeSlices and collate them by user, task, and/or day.

Our model looks like this (simplified):

class TimeSlice(models.Model):
    task = models.ForeignKey(Task)
    user = models.ForeignKey(User)
    begin = models.DateTimeField(default=datetime.datetime.now)
    duration = models.PositiveIntegerField(null=True, blank=True) # Num. of seconds
    note = models.TextField(null=True, blank=True)

Answer 1

Sounds like you want the aggregation functionality which is coming in Django 1.1. It's already available in recent checkouts from trunk.

See here for an explanation.