Putting a Date object into MongoDB, getting back a float when querying with pymongo

Question

I'm adding a date value into a MongoDB collection as part of a map-reduce call:

day = Date.UTC(this.time.getFullYear(), this.time.getMonth(), this.time.getDate());
emit({ user : this.user, day : day }, { count : 1 });

When I later query this collection in the Mongo shell I see:

{ "_id" : { "user" : "assaf", "day" : 1331769600000 }, "value" : { "count" : 15 } }
{ "_id" : { "user" : "assaf", "day" : 1331856000000 }, "value" : { "count" : 57 } }

Somehow the date looks like an integer - I guess it's some timestamp representation. If I do this:

PRIMARY> new Date(db.my_collection.find()[0]["_id"]["day"])

I get back the correct date:

ISODate("2012-03-19T00:00:00Z")

My question is how to do the same in pymongo. If I run any query on the above collection, pymongo returns documents in which the day value as a float type with the same value as the timestamp:

dict: {u'_id': {u'user': u'ariel', u'day': 1332115200000.0}, u'value': {u'count': 99.0}}

How do I turn this timestamp into a Python datetime?

Answer 1

Looks like milliseconds since epoch (1 Jan 1970):

>>> from __future__ import division
>>> dict = {u'_id': {u'user': u'ariel', u'day': 1332115200000.0}, u'value': {u'count': 99.0}}
>>> datetime.datetime.utcfromtimestamp(dict['_id']['day'] / 1000.0)
datetime.datetime(2012, 3, 19, 0, 0)
>>>

UPDATE: Added division check from first comment.