import random, string
ID_FIELD_LENGTH = 16
class ModelWithRandomID(models.Model):
id = models.CharField(primary_key=True, max_length=ID_FIELD_LENGTH)
def save(self):
if not self.id:
self.id = random_id(ID_FIELD_LENGTH)
super(ModelWithRandomID, self).save()
class Meta:
abstract = True
# alphabet will become our base-32 character set:
alphabet = string.lowercase + string.digits
# We must remove 4 characters from alphabet to make it 32 characters long. We want it to be 32
# characters long so that we can use a whole number of random bits to index into it.
for loser in 'l1o0': # Choose to remove ones that might be visually confusing
i = alphabet.index(loser)
alphabet = alphabet[:i] + alphabet[i+1:]
def byte_to_base32_chr(byte):
return alphabet[byte & 31]
def random_id(length):
# Can easily be converted to use secure random when available
# see http://www.secureprogramming.com/?action=view&feature=recipes&recipeid=20
random_bytes = [random.randint(0, 0xFF) for i in range(length)]
return ''.join(map(byte_to_base32_chr, random_bytes))
Comments
If you accidently create a new ID that is the same as an existing one, django will overwrite the existing one with your new data. At least, that's how it used to be.
This situation is unlikely for smaller databases but would become more and more likely for bigger ones.
Perhaps it would be better to keep an
AutoFieldprimary key, and use another field for your random ID, that isn't the primary key but hasunique=Trueand regenerates the random ID if there are problems saving.#
Also: UUIDs are in, and python even has a module for it :-)
#
awesome; thank you!
#