Django get_or_create trying to create row, throwing IntegrityError? - django

Question

I'm working in Django 1.7 and Postgres, and using the ORM to create some new rows. I am using get_or_create as follows:
p, created = Practice.objects.get_or_create(
code=row[1],
name=row[2],
address1=row[3],
address2=row[4],
address3=row[5],
address4=row[6],
postcode=row[7]
)
But when I try to run this I get:
django.db.utils.IntegrityError: duplicate key value violates unique constraint
DETAIL: Key (code)=(A82057) already exists
What's this about? I thought the point of get_or_create was to only try to create new rows if they didn't already exist.
My model looks like this:
class Practice(TimeStampedModel):
code = models.CharField(max_length=6, primary_key=True, db_index=True)
name = models.CharField(max_length=200)
address1 = models.CharField(max_length=200, null=True, blank=True)
address2 = models.CharField(max_length=200, null=True, blank=True)
address3 = models.CharField(max_length=200, null=True, blank=True)
address4 = models.CharField(max_length=200, null=True, blank=True)
postcode = models.CharField(max_length=9, null=True, blank=True)
def __str__(self):
return self.name
class Meta:
app_label = 'frontend'
ordering = ['name']
Is it something to do with the fact that I've set a manual primary key? I can't see anything in the Django docs about this restriction.

Answer 1

get_or_create attempts to do a get with all of the parameters you pass, not just the PK. So if there is an object with a matching PK but a different postcode, for example, the get will fail so a create will be attempted - but, since you have a manual PK, it will try to create a duplicate one using the data you have passed.
Generally speaking using a non-autoincrementing PK is a bad idea. But if you are just trying to look up against the PK only, use the defaults argument:
p, created = Practice.objects.get_or_create(
code=row[1],
defaults={
'name': row[2],
'address1': row[3],
'address2': row[4],
'address3': row[5],
'address4': row[6],
'postcode': row[7]
})