MongoDB: order positions by timedelta between CLOSE_DATE and NOW

gistfile1.txt

from pymongo import MongoClient
from bson import SON
from time import mktime
from datetime import datetime, timedelta
from time import mktime
from pprint import pprint
from random import randrange, randint, choice


def seed_position_times(db):
    for position in db.positions.find():
        open_day = randrange(
            mktime(now.timetuple()),
            mktime(month_later.timetuple())
        )
        close_day = randrange(
            open_day,
            open_day + randint(24*60*60, 24*60*60*3)
        )
        position['open_starts_on'] = datetime.fromtimestamp(open_day)
        position['open_ends_on'] = datetime.fromtimestamp(close_day)
        position['status'] = choice(['ok', 'closed'])
        db.positions.save(position)


now = datetime.utcnow()
month_later = now + timedelta(days=30)

db = MongoClient().inshipping
# seed_position_times(db)

filters = {
    'status': 'ok',
    'open_starts_on': {'$gt': now}
}

result = db.positions.aggregate([
    {'$match':
         filters
    },
    {'$project': {
        'timedelta': {'$subtract': ['$open_ends_on', now]},
        'open_starts_on': '$open_starts_on',
    }},
    {'$sort': SON([
        ('timeDelta', 1),
    ])},
    {'$project': {
        'open_starts_on': '$open_starts_on',
    }},
])

print(now)
print('---')
pprint([item['open_starts_on'] for item in result['result']][0:10])