Question

So, I have a model that looks like this:

class Names(models.Model):
  name = models.CharField()

class Entries(models.Model):
  name = models.ForeignKey(Names)
  date = models.DateField()

The data looks something like this

name   |  date

name_1 |  2011-06-01
name_2 |  2011-03-01
name_3 |  2011-02-01
name_1 |  2010-06-01
name_2 |  2010-03-02
name_3 |  2010-02-01
name_4 |  2009-07-01

I want to query Entries with a date, and get each name record once where it is less than or equal to the date. So, I want this returned if I query with date 2011-06-01:

name_1 | 2011-06-01
name_2 | 2011-03-01
name_3 | 2011-02-01
name_4 | 2009-07-01

My assumption is this would do the trick:

date = datetime(year=2011, month=06, day=01)
results = Entries.objects.filter(date__lte=date).order_by('date').distinct('name')

But, I keep getting duplicate name entries in there.

Any tips, friends?

Edit: The solution was provided by Gareth Rees. I had to modify it slightly, but it looks like this:

sql = """myapp_entries.id = (SELECT E.id from myapp_entries AS E
                             WHERE E.name_id = myapp_names.id AND
                             '"""+date.strftime('%Y-%m-%d')+"""'
                             ORDER BY E.date DESC
                             LIMIT 1)'''

results = (Entries.objects
           .extra(where = [sql])
           .filter(date__lte = date)
           .order_by('name', 'date'))

Answer 1

Here are a couple of approaches, depending on how you need the results.

(1) If you don’t mind getting the results in dictionary form (rather than Entries objects) then you can use annotate combined with values to group the results. In this case the query you want would be something like this:

from django.db.models import Min
results = (Entries.objects
           .filter(date__lte = date)
           .values('name')
           .annotate(date = Min('date'))
           .order_by('name', 'date'))

(2) If you need to get the results back as Entries objects, then you can use extra to pick only entries which are the earliest among all entries with the same name:

sql = '''id = (SELECT E.id FROM myapp_entries AS E
                WHERE E.name_id = myapp_entries.name_id
                ORDER BY E.date
                LIMIT 1)'''
results = (Entries.objects
           .extra(where = [sql])
           .filter(date__lte = date)
           .order_by('name', 'date'))