[SOLVED] Unique Object_id from timestamp with Python

Unique Object_id from timestamp with Python

I am creating a collection in MongoDB using pymongo. The elements of the collection already have a creation_date and I would like to use that to create the _id field. Since there is the possibility that more elements share the same creation_date, how can I create unique Object_id from that field?

bson has a function: bson.ObjectId.from_datetime(timestamp) which warns about the non uniqueness of the generated id. Is there a way to add some randomness to it such that different object_ids are generated from the same date?

Solution

The bson.ObjectId return a 24 hex digits variable which the first 8 digits is depends on the time and the remaining digits are some random number. If you create an id by using bson.ObjectId.from_datetime(args) function, it returns a 24 hex digits variable which last 16 digits are zero. By adding a random number to last part you can make it unique.

For generating random number you can use bson.ObjectId itself (last 16 digits).

ps: bson.ObjectId() returns a class instance and you should convert it into string.

import time
from datetime import datetime
import bson

def uniqueTimeIdCreator(_time = time.time()):

    date = datetime.fromtimestamp(_time)
    timeId = str(bson.ObjectId.from_datetime(date))
    uniqueId = str(bson.ObjectId())

    return timeId[0:8] + uniqueId[8:]

specTime = time.time()

id1 = uniqueTimeIdCreator(specTime)
id2 = uniqueTimeIdCreator(specTime)

print(id1)
print(id2)

You can check that id1 and id2 are the same timestamp constructing datetimes back from the strings:

t1 = bson.ObjectId(id1).generation_time
t2 = bson.ObjectId(id2).generation_time

print(t1)
print(t2)