While subclassing db.models.Model, sometimes it's essential to add extra checks/constraints.
For example, I have an Event model with start_date and end_date: I want to add validation into the fields or the model so that end_date > start_date.
At least I know this can be done outside the models.Model inside the ModelForm validation. But how to attach to the fields and the models.Model?
I would not put constraints like these in the save method, it's too late. Raising an exception there, doesn't help the user who entered the data in the wrong way, because it will end up as a 500 and the user won't get the form with errors back etc.
You should really check for this in the Forms/ModelForms clean method and raise a ValidationError, so form.is_valid() returns false and you can send the errors in the form back to the user for correction.
Also note that since version 1.2, Django has had Model Validation.
It would look something like this:
class Foo(models.Model):
# ... model stuff...
def clean(self):
if self.start_date > self.end_date:
raise ValidationError('Start date is after end date')
As of Django 2.2, database level constraints are supported:
from django.db import models
from django.db.models import CheckConstraint, Q, F
class Event(models.Model):
start_date = models.DatetimeField()
end_date = models.DatetimeField()
class Meta:
constraints = [
CheckConstraint(
check = Q(end_date__gt=F('start_date')),
name = 'check_start_date',
),
]
Do it inside your save method of your model:
def save(self, *args, **kwargs):
if(self.end_date > self.start_date):
super(Foo, self).save(*args, **kwargs)
else:
raise Exception, "end_date should be greater than start_date"
As #stefanw says, it's better user experience to check in the form's clean method.
This is enough if you're very sure that there isn't, and never will be, another way to change the value. But since you can rarely be sure of that, if database consistency is important, you can add another check (in addition to the form), one of:
The easier and database-independent way is in the model's save method as #umnik700 said. Note that this still doesn't prevent other users of the database (another app, or the admin interface) from creating an inconsistent state.
To be 'completely' sure the database is consistent, you can add a database level constraint. E.g. you can create a migration with RunSQL and SQL, something like (not tested):
migrations.RunSQL('ALTER TABLE app_event ADD CONSTRAINT chronology CHECK (start_date > end_date);')
(Not tested). This may be database dependent, which is a downside of course.
In your example, it's probably not worth it (incorrect start/end times just look a bit weird, but affect only the one inconsistent event), and you don't want manual schema changes. But it's useful in cases where consistency is critical.
EDIT: You can also just save the start time and the duration, instead of the start and end times.
As of today, both postgres 9.4 and MS SQL Server >= 2008 support check constraints in sql. On top of this, there is django issue 11964 which seems to be ready for review since yesterday, so hopefully we'll see this integrated into django 2. The project rapilabs/django-db-constraints seems to implement this too.
Summarizing the answers from before, here is a complete solution I used for a project:
from django.db import models
from django.db.models import CheckConstraint, Q, F
from django.utils.translation import gettext_lazy as _
class Event(models.Model):
start_date = models.DatetimeField()
end_date = models.DatetimeField()
class Meta:
constraints = [
# Ensures constraint on DB level, raises IntegrityError (500 on debug=False)
CheckConstraint(
check=Q(end_date__gt=F('start_date')), name='check_start_date',
),
]
def clean(self):
# Ensures constraint on model level, raises ValidationError
if self.start_date > self.end_date:
# raise error for field
raise ValidationError({'end_date': _('End date cannot be smaller then start date.')})
Too bad there is no django.core.validators that can handle this :(
Related
from django.db import models
from djago.db.models import F, Q
class(models.Model):
order_date = models.DateField()
class OrderLine(models.Model):
order = models.ForeignKeyField(Order)
loading_date = models.DateField()
class Meta:
constraints = [
models.CheckConstraint(check=Q(loading_date__gte=F("order__order_date")), name="disallow_backdated_loading")
I want to make sure always orderline loading_date is higher than orderdate
A CHECK constraint can span over a column, or over a table, but not over multiple tables, so this is not possible through a CHECK constraint.
Some databases allow to define triggers. These triggers run for example when a records is created/updated, and can run SQL queries, and decide to reject the creation/update based on such queries, but currently, the Django ORM does not support that.
One could also work with a composite primary key of an id and the creation date of the order, in which case the create timestamp is thus stored in the OrderLine table, and thus one can implement a check at the table level, but Django does not support working with composite primary keys for a number of reasons.
Therefore, besides running raw SQL, for example with a migration file that has a RunSQL operation [Django-doc], but this will likely be specific towards a database.
Therefore probably the most sensical check is to override the model clean() method [Django-doc]. Django however does not run the .clean() method before saving an object in the database, this is only done by ModelForms, and ModelAdmins. We can thus add a check with:
from django.core.exceptions import ValidationError
class OrderLine(models.Model):
order = models.ForeignKeyField(
Order,
on_delete=models.CASCADE
)
loading_date = models.DateField()
def clean(self):
if self.loading_date < self.order.order_date:
raise ValidationError('Can not load before ordering')
return super().clean()
This is an extension from my post here preventing crud operations on django model
A short into to the problem , im currently using a package called django-river to implement a workflow system in my application. The issue is that they do not have a predefined 'start' , 'dropped' , 'completed' state. Their states are stored as a django model instance. This would mean that my application is unable to programmatically differentiate between the states. Therefore , the labels of these states has to be hardcoded into my program (Or does it? Maybe someone has a solution to this?)
Suppose that there is no solution to the issue other than hardcoding the states into my application , this would mean that i would have to prevent users from updating , or deleting these states that i have pre created initially.
My idea is to have a form of validation check within the django model's save method . This check would check that the first 3 instances of the State model is always start , deactivated and completed and in the same order. This would prevent the check from passing through whenever a user trys to change items at the ORM level.
However , it would seem that there is 2 issues with this:
I believe django admin doesn't run the model class save method
Someone is still able to change the states as long as the way they changed it does not pass through the save() method. AKA from the DB SQL commands
Although it is unlikely to happen , changing the name would 'break' my application and therefore i wish to be very sure that no one can edit and change these 3 predefined states.
Is there a fool proof way to do this?
My idea is to have a form of validation check within the django model's save method.
if i understand your description, maybe you can just override the save() function of your model like so:
class MyModel(models.Model):
[..]
def save(self, *args, **kwargs):
# Put your logic here ..
super(MyModel, self).save(*args, **kwargs)
I got the answer from django documentation
from django.core.exceptions import ValidationError
from django.utils.translation import gettext_lazy as _
def validate_even(value):
if value % 2 != 0:
raise ValidationError(
_('%(value)s is not an even number'),
params={'value': value},
)
You can add this to a model field via the field’s validators argument:
from django.db import models
class MyModel(models.Model):
even_field = models.IntegerField(validators=[validate_even])
FYI: It is not really mandatory to use gettext_lazy and you can use just message as follows
from django.core.exceptions import ValidationError
def validate_even(value):
if value % 2 != 0:
raise ValidationError(
('%(value)s is not an even number'),
params={'value': value},
)
Is it possible to manually specify the set of related object to show in an inline, where no foreign key relation exists?
# Parent
class Diary(models.Model):
day = models.DateField()
activities = models.TextField()
# Child
class Sleep(models.Model):
start_time = models.DateTimeField()
end_time = models.DateTimeField()
class SleepInline(admin.TabularInline):
model=Sleep
def get_queryset(self, request):
# Return all Sleep objects where start_time and end_time are within Diary.day
return Sleep.objects.filter(XXX)
class DiaryAdmin(admin.ModelAdmin):
inlines = (SleepInline, )
I want my Diary model admin to display an inline for Sleep models that have start_time equal to the same day as Diary.day. The problem is that the Sleep model does not have a ForeignKey to Diary (instead, the relation is implicit by the use of dates).
Using the above, Django immediately complains that
<class 'records.admin.SleepInline'>: (admin.E202) 'records.Sleep' has no ForeignKey to 'records.Diary'.
How can I show the relevant Sleep instances as inlines on the Diary admin page?
There is no getting around the fact that Django admin inlines are built around ForeignKey fields (or ManyToManyField, OneToOneField). However, if I understand your goal, it's to avoid having to manage "date integrity" between your Diary.day and Sleep.start_time fields, i.e., the redundancy in a foreign key relation when that relation is really defined by Diary.day == Sleep.start_time.date()
A Django ForiegnKey field has a to_field property that allows the FK to index a column besides id. However, as you have a DateTimeField in Sleep and a DateField in Diary, we'll need to split that DateTimeField up. Also, a ForeignKey has to relate to something unique on the "1" side of the relation. Diary.day needs to be set unique=True.
In this approach, your models look like
from django.db import models
# Parent
class Diary(models.Model):
day = models.DateField(unique=True)
activities = models.TextField()
# Child
class Sleep(models.Model):
diary = models.ForeignKey(Diary, to_field='day', on_delete=models.CASCADE)
start_time = models.TimeField()
end_time = models.DateTimeField()
and then your admin.py is just
from django.contrib import admin
from .models import Sleep, Diary
class SleepInline(admin.TabularInline):
model=Sleep
#admin.register(Diary)
class DiaryAdmin(admin.ModelAdmin):
inlines = (SleepInline, )
Even though Sleep.start_time no longer has a date, the Django Admin is quite what you'd expect, and avoids "date redundancy":
Thinking ahead to a more real (and problematic) use case, say every user can have 1 Diary per day:
class Diary(models.Model):
user = models.ForeignKey(User)
day = models.DateField()
activities = models.TextField()
class Meta:
unique_together = ('user', 'day')
One would like to write something like
class Sleep(models.Model):
diary = models.ForeignKey(Diary, to_fields=['user', 'day'], on_delete=models.CASCADE)
However, there's no such feature in Django 1.11, nor can I find any serious discussion of adding that. Certainly composite foreign keys are allowed in Postgres and other SQL DBMS's. I get the impression from the Django source they're keeping their options open: https://github.com/django/django/blob/stable/1.11.x/django/db/models/fields/related.py#L621 hints at a future implementation.
Finally, https://pypi.python.org/pypi/django-composite-foreignkey looks interesting at first, but doesn't create "real" composite foreign keys, nor does it work with Django's admin.
Let me start by showing you the drawbacks of your logic:
When adding a foreign key, there are 2 operations, that are uncommon that require adjusting the relation: creating a new sleep object and updating the times on the sleep object.
When not using a foreign key, each time a diary is requested the lookup for the corresponding Sleep object(s) needs to be done. I'm assuming reading diaries is much more common then alterations of sleep objects, as it will be in most projects out there.
The additional drawback as you've noticed, is that you cannot use relational features. InlineAdmin is a relational feature, so as much as you say "making the admin work", it is really that you demand a hammer to unscrew a bolt.
But...the admin makes use of ModelForm. So if you construct the form with a formset (which cannot be a an inline formset for the same reason) and handle saving that formset yourself, it should be possible. The whole point of InlineFormset and InlineAdmin is to make generation of formsets from related models easier and for that it needs to know the relation.
And finally, you can add urls and build a custom page, and when extending the admin/base.html template, you will have access to the layout and javascript components.
You can achieve this, using nested_admin library.
So in myapp/admin.py:
from nested_admin.nested import NestedStackedInline, NestedModelAdmin
class SleepInline(NestedStackedInline):
model = Sleep
class DiaryAdmin(NestedModelAdmin):
inlines = [SleepInline]
I need to add an autoinc field that is not the primary key. I am in the process of migrating a very large production database that uses autoincrementing fields to models.UUIDField. I have been doing a piecewise migration, and all of my relationships are now duplicated with both field types. I'm ready to make the primary key swap, but unfortunately I still need to keep the auto incrementing integer field for old clients as it becomes deprecated.
Since django will not allow me to define an autofield with primary_key=False (even though this is fully supported at the db layer), i'm looking for a simple solution. My initial strategy would be to simply change the field to models.BigIntegerField('GUID', db_index=True, null=True, unique=True) and then manually set the default nextval('my_guid_seq'::regclass) using migrations.RunSQL. So far so good, except not. It turns out, because of my null=True declaration, django at the ORM layer is taking over and inserting null which will not allow defaults at the database layer to do it's job.
The core developers are fast to reject this request because of bad design, which I most definetly agree with, but there are very valid use cases such as this. https://code.djangoproject.com/ticket/8576
I am a very weak django developer so I don't want to get in the weeds metaprogramming at the ORM layer. This is by definition a hack, so i'm looking for the least complex, creative solution that gets me around this limitation
You could subclass AutoField and override the _check_primary_key method.
from django.db.models.fields import AutoField
from django.db.models.fields import checks
class AutoFieldNonPrimary(AutoField):
def _check_primary_key(self):
if self.primary_key:
return [
checks.Error(
"AutoFieldNonPrimary must not set primary_key=True.",
obj=self,
id="fields.E100",
)
]
else:
return []
See AutoField source code here
Edit: Updated Link
I know, changing the primary key to UUID is such a pain.Hence the simple and better solution that I think of is to add another integer field that is auto-incrementing in nature.
Here is my solution:
class ModelName(models.Model):
auto_inc_id = models.IntegerField()
Then override the save model:
def save(self, *args, **kwargs):
self.object_list = ModelName.objects.order_by('auto_inc_id')
if len(self.object_list) == 0: # if there are no objects
self.auto_inc_id = 1
else:
self.auto_inc_id = self.object_list.last().auto_inc_id + 1
super(ModelName, self).save()
Couldn't format this as a comment, but modifying #Abhimanyu's answer to make the save method more concise (and issue only one query). Same model property:
class ModelName(models.Model):
auto_inc_id = models.IntegerField()
And here's the save method on the model:
def save(self, *args, **kwargs):
self.auto_inc_id = ModelName.objects.all().count() + 1
super(ModelName, self).save()
class dbview(models.Model):
# field definitions omitted for brevity
class Meta:
db_table = 'read_only_view'
def main(request):
result = dbview.objects.all()
Caught an exception while rendering: (1054, "Unknown column 'read_only_view.id' in 'field list'")
There is no primary key I can see in the view. Is there a workaround?
Comment:
I have no control over the view I am accessing with Django. MySQL browser shows columns there but no primary key.
When you say 'I have no control over the view I am accessing with Django. MySQL browser shows columns there but no primary key.'
I assume you mean that this is a legacy table and you are not allowed to add or change columns?
If so and there really isn't a primary key (even a string or non-int column*) then the table hasn't been set up very well and performance might well stink.
It doesn't matter to you though. All you need is a column that is guaranteed to be unique for every row. Set that to be 'primary_key = True in your model and Django will be happy.
There is one other possibility that would be problemmatic. If there is no column that is guaranteed to be unique then the table might be using composite primary keys. That is - it is specifying that two columns taken together will provide a unique primary key. This is perfectly valid relational modelling but unfortunatly unsupported by Django. In that case you can't do much besides raw SQL unless you can get another column added.
I have this issue all the time. I have a view that I can't or don't want to change, but I want to have a page to display composite information (maybe in the admin section). I just override the save and raise a NotImplementedError:
def save(self, **kwargs):
raise NotImplementedError()
(although this is probably not needed in most cases, but it makes me feel a bit better)
I also set managed to False in the Meta class.
class Meta:
managed = False
Then I just pick any field and tag it as the primary key. It doesn't matter if it's really unique with you are just doing filters for displaying information on a page, etc.
Seems to work fine for me. Please commment if there are any problems with this technique that I'm overlooking.
If there really is no primary key in the view, then there is no workaround.
Django requires each model to have exactly one field primary_key=True.
There should have been an auto-generated id field when you ran syncdb (if there is no primary key defined in your model, then Django will insert an AutoField for you).
This error means that Django is asking your database for the id field, but none exists. Can you run django manage.py dbshell and then DESCRIBE read_only_view; and post the result? This will show all of the columns that are in the database.
Alternatively, can you include the model definition you excluded? (and confirm that you haven't altered the model definition since you ran syncdb?)
I know this post is over a decade old, but I ran into this recently and came to SO looking for a good answer. I had to come up with a solution that addresses the OP's original question, and, additionally, allows for us to add new objects to the model for unit testing purposes, which is a problem I still had with all of the provided solutions.
main.py
from django.db import models
def in_unit_test_mode():
"""some code to detect if you're running unit tests with a temp SQLite DB, like..."""
import sys
return "test" in sys.argv
"""You wouldn't want to actually implement it with the import inside here. We have a setting in our django.conf.settings that tests to see if we're running unit tests when the project starts."""
class AbstractReadOnlyModel(models.Model):
class Meta(object):
abstract = True
managed = in_unit_test_mode()
"""This is just to help you fail fast in case a new developer, or future you, doesn't realize this is a database view and not an actual table and tries to update it."""
def save(self, *args, **kwargs):
if not in_unit_test_mode():
raise NotImplementedError(
"This is a read only model. We shouldn't be writing "
"to the {0} table.".format(self.__class__.__name__)
)
else:
super(AbstractReadOnlyModel, self).save(*args, **kwargs)
class DbViewBaseModel(AbstractReadOnlyModel):
not_actually_unique_field = IntegerField(primary_key=True)
# the rest of your field definitions
class Meta:
db_table = 'read_only_view'
if in_unit_test_mode():
class DbView(DbViewBaseModel):
not_actually_unique_field = IntegerField()
"""This line removes the primary key property from the 'not_actually_unique_field' when running unit tests, so Django will create an AutoField named 'id' on the table it creates in the temp DB that it creates for running unit tests."""
else:
class DbView(DbViewBaseModel):
pass
class MainClass(object):
#staticmethod
def main_method(request):
return DbView.objects.all()
test.py
from django.test import TestCase
from main import DbView
from main import MainClass
class TestMain(TestCase):
#classmethod
def setUpTestData(cls):
cls.object_in_view = DbView.objects.create(
"""Enter fields here to create test data you expect to be returned from your method."""
)
def testMain(self):
objects_from_view = MainClass.main_method()
returned_ids = [object.id for object in objects_from_view]
self.assertIn(self.object_in_view.id, returned_ids)