#34792: Creating and saving a model using a custom primary key field can yield a
bad "id" value on the instance
-------------------------------------+-------------------------------------
Reporter: mike w | Owner: nobody
Type: Bug | Status: new
Component: Database layer | Version: 4.2
(models, ORM) |
Severity: Normal | Resolution:
Keywords: field, custom | Triage Stage:
fields | Unreviewed
Has patch: 0 | Needs documentation: 0
Needs tests: 0 | Patch needs improvement: 0
Easy pickings: 0 | UI/UX: 0
-------------------------------------+-------------------------------------
Changes (by mike w):
* type: Uncategorized => Bug
Old description:
> **Summary**: In some situations, creating and saving a model which uses a
> custom primary key can cause a wrong ".id" value to be set on the
> instance in memory.
>
> **Detail**: I am using a custom primary key Field type which inherits
> from AutoField, [https://github.com/mik3y/django-spicy-id django-spicy-
> id]. Essentially what this Field does is translate the underlying integer
> data in the database, to and from a string (with a static prefix) at the
> Django level.
>
> It does this by implementing [https://github.com/mik3y/django-spicy-
> id/blob/ce8cf54376f4cf0f82de9f385afaa70165eff2e3/django_spicy_id/fields.py#L147-L150
> from_db_value() and friends] in the Field implementation.
>
> This might not be to everyone's taste, but so far I think it is
> completely within the acceptable functionality & contract of custom
> Fields.
>
> I noticed that in some environments, particularly in Django pre-4.0,
> `save()` on a new instance would cause the instance's `.id` field to be
> the underlying number - and ''not'' the expected string value, which the
> custom Field implements. Example:
>
> {{{
> class MyModel(models.Model):
> id = SpicyBigAutoField("ex", primary_key=True, randomize=True)
> }}}
>
> {{{
> >>> o = MyModel()
> >>> o.save()
> >> o.id
> 3735928559 # yikes! we should never get a number here.
> >>> o.refresh_from_db()
> >>> o.id
> 'ex_deadbeef`
> }}}
>
> I believe the culprit is
> [https://github.com/django/django/blob/fbd16438f46bc2128926958ad24331da5d1b406f/django/db/models/sql/compiler.py#L1825-L1847
> this block] in django.db.models.sql.compiler, specifically the last
> condition where self.connection.ops.last_insert_id is called and returned
> without modification.
>
> In turn, the parent caller
> [https://github.com/django/django/blob/fbd16438f46bc2128926958ad24331da5d1b406f/django/db/models/base.py#L1057-L1063
> django.db.models.base.Model::_save_table()] will blindly setattr this
> value on the instance.
>
> It seems like a possible fix could be to flow the "last_insert_id" case
> through the Field's from_db_value(); in most cases this is surely a no-
> op, but.. not in this case.
>
> (I'd happy to prepare a patch and/or test if offered a little guidance.)
> Thanks!
New description:
**Summary**: In some situations, creating and saving a model which uses a
custom primary key can cause a wrong ".id" value to be set on the instance
in memory.
**Detail**: I am using a custom primary key Field type which inherits from
AutoField, [https://github.com/mik3y/django-spicy-id django-spicy-id].
Essentially what this Field does is translate the underlying integer data
in the database, to and from a string (with a static prefix) at the Django
level.
It does this by implementing [https://github.com/mik3y/django-spicy-
id/blob/ce8cf54376f4cf0f82de9f385afaa70165eff2e3/django_spicy_id/fields.py#L147-L150
from_db_value() and friends] in the Field implementation.
This might not be to everyone's taste, but so far I think it is completely
within the acceptable functionality & contract of custom Fields.
I noticed that in some environments, particularly in Django pre-4.0,
`save()` on a new instance would cause the instance's `.id` field to be
the underlying number - and ''not'' the expected string value, which the
custom Field implements. Example:
{{{
class MyModel(models.Model):
id = SpicyBigAutoField("ex", primary_key=True, randomize=True)
}}}
{{{
>>> o = MyModel()
>>> o.save()
>>> o.id
3735928559 # yikes! we should never get a number here.
>>> o.refresh_from_db()
>>> o.id
'ex_deadbeef`
}}}
I believe the culprit is
[https://github.com/django/django/blob/fbd16438f46bc2128926958ad24331da5d1b406f/django/db/models/sql/compiler.py#L1825-L1847
this block] in django.db.models.sql.compiler, specifically the last
condition where self.connection.ops.last_insert_id is called and returned
without modification.
In turn, the parent caller
[https://github.com/django/django/blob/fbd16438f46bc2128926958ad24331da5d1b406f/django/db/models/base.py#L1057-L1063
django.db.models.base.Model::_save_table()] will blindly setattr this
value on the instance.
It seems like a possible fix could be to flow the "last_insert_id" case
through the Field's from_db_value(); in most cases this is surely a no-op,
but.. not in this case.
(I'd happy to prepare a patch and/or test if offered a little guidance.)
Thanks!
--
--
Ticket URL: <https://code.djangoproject.com/ticket/34792#comment:1>
Django <https://code.djangoproject.com/>
The Web framework for perfectionists with deadlines.
--
You received this message because you are subscribed to the Google Groups
"Django updates" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/django-updates/0107018a1f4027ed-1da1fb9d-3f2f-476a-bcd0-a8183ad4aad6-000000%40eu-central-1.amazonses.com.
