Model instance reference

This document describes the details of the Model API. It builds on the material presented in the model and database query guides, so you’ll probably want to read and understand those documents before reading this one.

Throughout this reference we’ll use the example Weblog models presented in the database query guide.

Creating objects

To create a new instance of a model, instantiate it like any other Python class:

class Model(**kwargs)

The keyword arguments are the names of the fields you’ve defined on your model. Note that instantiating a model in no way touches your database; for that, you need to save().

Customizing model loading

classmethod Model.from_db(db, field_names, values)

The from_db() method can be used to customize model instance creation when loading from the database.

The db argument contains the database alias for the database the model is loaded from, field_names contains the names of all loaded fields, and values contains the loaded values for each field in field_names. The field_names are in the same order as the values. If all of the model’s fields are present, then values are guaranteed to be in the order __init__() expects them. That is, the instance can be created by cls(*values). If any fields are deferred, they won’t appear in field_names. In that case, assign a value of django.db.models.DEFERRED to each of the missing fields.

In addition to creating the new model, the from_db() method must set the adding and db flags in the new instance’s _state attribute.

Below is an example showing how to record the initial values of fields that are loaded from the database:

from django.db.models import DEFERRED

@classmethod
def from_db(cls, db, field_names, values):
    # Default implementation of from_db() (subject to change and could
    # be replaced with super()).
    if len(values) != len(cls._meta.concrete_fields):
        values = list(values)
        values.reverse()
        values = [
            values.pop() if f.attname in field_names else DEFERRED
            for f in cls._meta.concrete_fields
        ]
    instance = cls(*values)
    instance._state.adding = False
    instance._state.db = db
    # customization to store the original field values on the instance
    instance._loaded_values = dict(zip(field_names, values))
    return instance

def save(self, *args, **kwargs):
    # Check how the current values differ from ._loaded_values. For example,
    # prevent changing the creator_id of the model. (This example doesn't
    # support cases where 'creator_id' is deferred).
    if not self._state.adding and (
            self.creator_id != self._loaded_values['creator_id']):
        raise ValueError("Updating the value of creator isn't allowed")
    super().save(*args, **kwargs)

The example above shows a full from_db() implementation to clarify how that is done. In this case it would of course be possible to use super() call in the from_db() method.

Refreshing objects from database

If you delete a field from a model instance, accessing it again reloads the value from the database:

>>> obj = MyModel.objects.first()
>>> del obj.field
>>> obj.field  # Loads the field from the database

Model.refresh_from_db(using=None, fields=None)

If you need to reload a model’s values from the database, you can use the refresh_from_db() method. When this method is called without arguments the following is done:

1. All non-deferred fields of the model are updated to the values currently present in the database.
2. Any cached relations are cleared from the reloaded instance.

Only fields of the model are reloaded from the database. Other database-dependent values such as annotations aren’t reloaded. Any @cached_property attributes aren’t cleared either.

The reloading happens from the database the instance was loaded from, or from the default database if the instance wasn’t loaded from the database. The using argument can be used to force the database used for reloading.

It is possible to force the set of fields to be loaded by using the fields argument.

For example, to test that an update() call resulted in the expected update, you could write a test similar to this:

def test_update_result(self):
    obj = MyModel.objects.create(val=1)
    MyModel.objects.filter(pk=obj.pk).update(val=F('val') + 1)
    # At this point obj.val is still 1, but the value in the database
    # was updated to 2. The object's updated value needs to be reloaded
    # from the database.
    obj.refresh_from_db()
    self.assertEqual(obj.val, 2)

Note that when deferred fields are accessed, the loading of the deferred field’s value happens through this method. Thus it is possible to customize the way deferred loading happens. The example below shows how one can reload all of the instance’s fields when a deferred field is reloaded:

class ExampleModel(models.Model):
    def refresh_from_db(self, using=None, fields=None, **kwargs):
        # fields contains the name of the deferred field to be
        # loaded.
        if fields is not None:
            fields = set(fields)
            deferred_fields = self.get_deferred_fields()
            # If any deferred field is going to be loaded
            if fields.intersection(deferred_fields):
                # then load all of them
                fields = fields.union(deferred_fields)
        super().refresh_from_db(using, fields, **kwargs)

Model.get_deferred_fields()

A helper method that returns a set containing the attribute names of all those fields that are currently deferred for this model.

Validating objects

There are three steps involved in validating a model:

1. Validate the model fields - Model.clean_fields()
2. Validate the model as a whole - Model.clean()
3. Validate the field uniqueness - Model.validate_unique()

All three steps are performed when you call a model’s full_clean() method.

When you use a ModelForm, the call to is_valid() will perform these validation steps for all the fields that are included on the form. See the ModelForm documentation for more information. You should only need to call a model’s full_clean() method if you plan to handle validation errors yourself, or if you have excluded fields from the ModelForm that require validation.

Model.full_clean(exclude=None, validate_unique=True)

This method calls Model.clean_fields(), Model.clean(), and Model.validate_unique() (if validate_unique is True), in that order and raises a ValidationError that has a message_dict attribute containing errors from all three stages.

The optional exclude argument can be used to provide a list of field names that can be excluded from validation and cleaning. ModelForm uses this argument to exclude fields that aren’t present on your form from being validated since any errors raised could not be corrected by the user.

Note that full_clean() will not be called automatically when you call your model’s save() method. You’ll need to call it manually when you want to run one-step model validation for your own manually created models. For example:

from django.core.exceptions import ValidationError
try:
    article.full_clean()
except ValidationError as e:
    # Do something based on the errors contained in e.message_dict.
    # Display them to a user, or handle them programmatically.
    pass

The first step full_clean() performs is to clean each individual field.

Model.clean_fields(exclude=None)

This method will validate all fields on your model. The optional exclude argument lets you provide a list of field names to exclude from validation. It will raise a ValidationError if any fields fail validation.

The second step full_clean() performs is to call Model.clean(). This method should be overridden to perform custom validation on your model.

Model.clean()

This method should be used to provide custom model validation, and to modify attributes on your model if desired. For instance, you could use it to automatically provide a value for a field, or to do validation that requires access to more than a single field:

import datetime
from django.core.exceptions import ValidationError
from django.db import models
from django.utils.translation import gettext_lazy as _

class Article(models.Model):
    ...
    def clean(self):
        # Don't allow draft entries to have a pub_date.
        if self.status == 'draft' and self.pub_date is not None:
            raise ValidationError(_('Draft entries may not have a publication date.'))
        # Set the pub_date for published items if it hasn't been set already.
        if self.status == 'published' and self.pub_date is None:
            self.pub_date = datetime.date.today()

Note, however, that like Model.full_clean(), a model’s clean() method is not invoked when you call your model’s save() method.

In the above example, the ValidationError exception raised by Model.clean() was instantiated with a string, so it will be stored in a special error dictionary key, NON_FIELD_ERRORS. This key is used for errors that are tied to the entire model instead of to a specific field:

from django.core.exceptions import NON_FIELD_ERRORS, ValidationError
try:
    article.full_clean()
except ValidationError as e:
    non_field_errors = e.message_dict[NON_FIELD_ERRORS]

To assign exceptions to a specific field, instantiate the ValidationError with a dictionary, where the keys are the field names. We could update the previous example to assign the error to the pub_date field:

class Article(models.Model):
    ...
    def clean(self):
        # Don't allow draft entries to have a pub_date.
        if self.status == 'draft' and self.pub_date is not None:
            raise ValidationError({'pub_date': _('Draft entries may not have a publication date.')})
        ...

If you detect errors in multiple fields during Model.clean(), you can also pass a dictionary mapping field names to errors:

raise ValidationError({
    'title': ValidationError(_('Missing title.'), code='required'),
    'pub_date': ValidationError(_('Invalid date.'), code='invalid'),
})

Finally, full_clean() will check any unique constraints on your model.

class Article(models.Model):
    ...
    def clean_fields(self, exclude=None):
        super().clean_fields(exclude=exclude)
        if self.status == 'draft' and self.pub_date is not None:
            if exclude and 'status' in exclude:
                raise ValidationError(
                    _('Draft entries may not have a publication date.')
                )
            else:
                raise ValidationError({
                    'status': _(
                        'Set status to draft if there is not a '
                        'publication date.'
                     ),
                })

Model.validate_unique(exclude=None)

This method is similar to clean_fields(), but validates all uniqueness constraints on your model instead of individual field values. The optional exclude argument allows you to provide a list of field names to exclude from validation. It will raise a ValidationError if any fields fail validation.

Note that if you provide an exclude argument to validate_unique(), any unique_together constraint involving one of the fields you provided will not be checked.

Saving objects

To save an object back to the database, call save():

Model.save(force_insert=False, force_update=False, using=DEFAULT_DB_ALIAS, update_fields=None)

If you want customized saving behavior, you can override this save() method. See Overriding predefined model methods for more details.

The model save process also has some subtleties; see the sections below.

Auto-incrementing primary keys

If a model has an AutoField — an auto-incrementing primary key — then that auto-incremented value will be calculated and saved as an attribute on your object the first time you call save():

>>> b2 = Blog(name='Cheddar Talk', tagline='Thoughts on cheese.')
>>> b2.id     # Returns None, because b2 doesn't have an ID yet.
>>> b2.save()
>>> b2.id     # Returns the ID of your new object.

There’s no way to tell what the value of an ID will be before you call save(), because that value is calculated by your database, not by Django.

For convenience, each model has an AutoField named id by default unless you explicitly specify primary_key=True on a field in your model. See the documentation for AutoField for more details.

The pk property

Model.pk

Regardless of whether you define a primary key field yourself, or let Django supply one for you, each model will have a property called pk. It behaves like a normal attribute on the model, but is actually an alias for whichever attribute is the primary key field for the model. You can read and set this value, just as you would for any other attribute, and it will update the correct field in the model.

Explicitly specifying auto-primary-key values

If a model has an AutoField but you want to define a new object’s ID explicitly when saving, define it explicitly before saving, rather than relying on the auto-assignment of the ID:

>>> b3 = Blog(id=3, name='Cheddar Talk', tagline='Thoughts on cheese.')
>>> b3.id     # Returns 3.
>>> b3.save()
>>> b3.id     # Returns 3.

If you assign auto-primary-key values manually, make sure not to use an already-existing primary-key value! If you create a new object with an explicit primary-key value that already exists in the database, Django will assume you’re changing the existing record rather than creating a new one.

Given the above 'Cheddar Talk' blog example, this example would override the previous record in the database:

b4 = Blog(id=3, name='Not Cheddar', tagline='Anything but cheese.')
b4.save()  # Overrides the previous blog with ID=3!

See How Django knows to UPDATE vs. INSERT, below, for the reason this happens.

Explicitly specifying auto-primary-key values is mostly useful for bulk-saving objects, when you’re confident you won’t have primary-key collision.

If you’re using PostgreSQL, the sequence associated with the primary key might need to be updated; see Manually-specifying values of auto-incrementing primary keys.

What happens when you save?

When you save an object, Django performs the following steps:

1. Emit a pre-save signal. The pre_save signal is sent, allowing any functions listening for that signal to do something.
2. Preprocess the data. Each field’s pre_save() method is called to perform any automated data modification that’s needed. For example, the date/time fields override pre_save() to implement auto_now_add and auto_now.

3. Prepare the data for the database. Each field’s get_db_prep_save() method is asked to provide its current value in a data type that can be written to the database.

   Most fields don’t require data preparation. Simple data types, such as integers and strings, are ‘ready to write’ as a Python object. However, more complex data types often require some modification.

   For example, DateField fields use a Python datetime object to store data. Databases don’t store datetime objects, so the field value must be converted into an ISO-compliant date string for insertion into the database.

4. Insert the data into the database. The preprocessed, prepared data is composed into an SQL statement for insertion into the database.
5. Emit a post-save signal. The post_save signal is sent, allowing any functions listening for that signal to do something.

How Django knows to UPDATE vs. INSERT

You may have noticed Django database objects use the same save() method for creating and changing objects. Django abstracts the need to use INSERT or UPDATE SQL statements. Specifically, when you call save() and the object’s primary key attribute does not define a default, Django follows this algorithm:

• If the object’s primary key attribute is set to a value that evaluates to True (i.e., a value other than None or the empty string), Django executes an UPDATE.
• If the object’s primary key attribute is not set or if the UPDATE didn’t update anything (e.g. if primary key is set to a value that doesn’t exist in the database), Django executes an INSERT.

If the object’s primary key attribute defines a default then Django executes an UPDATE if it is an existing model instance and primary key is set to a value that exists in the database. Otherwise, Django executes an INSERT.

The one gotcha here is that you should be careful not to specify a primary-key value explicitly when saving new objects, if you cannot guarantee the primary-key value is unused. For more on this nuance, see Explicitly specifying auto-primary-key values above and Forcing an INSERT or UPDATE below.

In Django 1.5 and earlier, Django did a SELECT when the primary key attribute was set. If the SELECT found a row, then Django did an UPDATE, otherwise it did an INSERT. The old algorithm results in one more query in the UPDATE case. There are some rare cases where the database doesn’t report that a row was updated even if the database contains a row for the object’s primary key value. An example is the PostgreSQL ON UPDATE trigger which returns NULL. In such cases it is possible to revert to the old algorithm by setting the select_on_save option to True.

Forcing an INSERT or UPDATE

In some rare circumstances, it’s necessary to be able to force the save() method to perform an SQL INSERT and not fall back to doing an UPDATE. Or vice-versa: update, if possible, but not insert a new row. In these cases you can pass the force_insert=True or force_update=True parameters to the save() method. Passing both parameters is an error: you cannot both insert and update at the same time!

It should be very rare that you’ll need to use these parameters. Django will almost always do the right thing and trying to override that will lead to errors that are difficult to track down. This feature is for advanced use only.

Using update_fields will force an update similarly to force_update.

Updating attributes based on existing fields

Sometimes you’ll need to perform a simple arithmetic task on a field, such as incrementing or decrementing the current value. The obvious way to achieve this is to do something like:

>>> product = Product.objects.get(name='Venezuelan Beaver Cheese')
>>> product.number_sold += 1
>>> product.save()

If the old number_sold value retrieved from the database was 10, then the value of 11 will be written back to the database.

The process can be made robust, avoiding a race condition, as well as slightly faster by expressing the update relative to the original field value, rather than as an explicit assignment of a new value. Django provides F expressions for performing this kind of relative update. Using F expressions, the previous example is expressed as:

>>> from django.db.models import F
>>> product = Product.objects.get(name='Venezuelan Beaver Cheese')
>>> product.number_sold = F('number_sold') + 1
>>> product.save()

For more details, see the documentation on F expressions and their use in update queries.

Specifying which fields to save

If save() is passed a list of field names in keyword argument update_fields, only the fields named in that list will be updated. This may be desirable if you want to update just one or a few fields on an object. There will be a slight performance benefit from preventing all of the model fields from being updated in the database. For example:

product.name = 'Name changed again'
product.save(update_fields=['name'])

The update_fields argument can be any iterable containing strings. An empty update_fields iterable will skip the save. A value of None will perform an update on all fields.

Specifying update_fields will force an update.

When saving a model fetched through deferred model loading (only() or defer()) only the fields loaded from the DB will get updated. In effect there is an automatic update_fields in this case. If you assign or change any deferred field value, the field will be added to the updated fields.

Deleting objects

Model.delete(using=DEFAULT_DB_ALIAS, keep_parents=False)

Issues an SQL DELETE for the object. This only deletes the object in the database; the Python instance will still exist and will still have data in its fields. This method returns the number of objects deleted and a dictionary with the number of deletions per object type.

For more details, including how to delete objects in bulk, see Deleting objects.

If you want customized deletion behavior, you can override the delete() method. See Overriding predefined model methods for more details.

Sometimes with multi-table inheritance you may want to delete only a child model’s data. Specifying keep_parents=True will keep the parent model’s data.

Pickling objects

When you pickle a model, its current state is pickled. When you unpickle it, it’ll contain the model instance at the moment it was pickled, rather than the data that’s currently in the database.

Other model instance methods

A few object methods have special purposes.

__str__()

Model.__str__()

The __str__() method is called whenever you call str() on an object. Django uses str(obj) in a number of places. Most notably, to display an object in the Django admin site and as the value inserted into a template when it displays an object. Thus, you should always return a nice, human-readable representation of the model from the __str__() method.

For example:

from django.db import models

class Person(models.Model):
    first_name = models.CharField(max_length=50)
    last_name = models.CharField(max_length=50)

    def __str__(self):
        return '%s %s' % (self.first_name, self.last_name)

__eq__()

Model.__eq__()

The equality method is defined such that instances with the same primary key value and the same concrete class are considered equal, except that instances with a primary key value of None aren’t equal to anything except themselves. For proxy models, concrete class is defined as the model’s first non-proxy parent; for all other models it’s simply the model’s class.

For example:

from django.db import models

class MyModel(models.Model):
    id = models.AutoField(primary_key=True)

class MyProxyModel(MyModel):
    class Meta:
        proxy = True

class MultitableInherited(MyModel):
    pass

# Primary keys compared
MyModel(id=1) == MyModel(id=1)
MyModel(id=1) != MyModel(id=2)
# Primary keys are None
MyModel(id=None) != MyModel(id=None)
# Same instance
instance = MyModel(id=None)
instance == instance
# Proxy model
MyModel(id=1) == MyProxyModel(id=1)
# Multi-table inheritance
MyModel(id=1) != MultitableInherited(id=1)

__hash__()

Model.__hash__()

The __hash__() method is based on the instance’s primary key value. It is effectively hash(obj.pk). If the instance doesn’t have a primary key value then a TypeError will be raised (otherwise the __hash__() method would return different values before and after the instance is saved, but changing the __hash__() value of an instance is forbidden in Python.

get_absolute_url()

Model.get_absolute_url()

Define a get_absolute_url() method to tell Django how to calculate the canonical URL for an object. To callers, this method should appear to return a string that can be used to refer to the object over HTTP.

For example:

def get_absolute_url(self):
    return "/people/%i/" % self.id

While this code is correct and simple, it may not be the most portable way to to write this kind of method. The reverse() function is usually the best approach.

For example:

def get_absolute_url(self):
    from django.urls import reverse
    return reverse('people.views.details', args=[str(self.id)])

One place Django uses get_absolute_url() is in the admin app. If an object defines this method, the object-editing page will have a “View on site” link that will jump you directly to the object’s public view, as given by get_absolute_url().

Similarly, a couple of other bits of Django, such as the syndication feed framework, use get_absolute_url() when it is defined. If it makes sense for your model’s instances to each have a unique URL, you should define get_absolute_url().

def get_absolute_url(self):
    return '/%s/' % self.name

It’s good practice to use get_absolute_url() in templates, instead of hard-coding your objects’ URLs. For example, this template code is bad:

<!-- BAD template code. Avoid! -->
<a href="/people/{{ object.id }}/">{{ object.name }}</a>

This template code is much better:

<a href="{{ object.get_absolute_url }}">{{ object.name }}</a>

The logic here is that if you change the URL structure of your objects, even for something small like correcting a spelling error, you don’t want to have to track down every place that the URL might be created. Specify it once, in get_absolute_url() and have all your other code call that one place.

Extra instance methods

In addition to save(), delete(), a model object might have some of the following methods:

Model.get_FOO_display()

For every field that has choices set, the object will have a get_FOO_display() method, where FOO is the name of the field. This method returns the “human-readable” value of the field.

For example:

from django.db import models

class Person(models.Model):
    SHIRT_SIZES = (
        ('S', 'Small'),
        ('M', 'Medium'),
        ('L', 'Large'),
    )
    name = models.CharField(max_length=60)
    shirt_size = models.CharField(max_length=2, choices=SHIRT_SIZES)

>>> p = Person(name="Fred Flintstone", shirt_size="L")
>>> p.save()
>>> p.shirt_size
'L'
>>> p.get_shirt_size_display()
'Large'

Model.get_next_by_FOO(**kwargs)

Model.get_previous_by_FOO(**kwargs)

For every DateField and DateTimeField that does not have null=True, the object will have get_next_by_FOO() and get_previous_by_FOO() methods, where FOO is the name of the field. This returns the next and previous object with respect to the date field, raising a DoesNotExist exception when appropriate.

Both of these methods will perform their queries using the default manager for the model. If you need to emulate filtering used by a custom manager, or want to perform one-off custom filtering, both methods also accept optional keyword arguments, which should be in the format described in Field lookups.

Note that in the case of identical date values, these methods will use the primary key as a tie-breaker. This guarantees that no records are skipped or duplicated. That also means you cannot use those methods on unsaved objects.

Other attributes

_state

Model._state

The _state attribute refers to a ModelState object that tracks the lifecycle of the model instance.

The ModelState object has two attributes: adding, a flag which is True if the model has not been saved to the database yet, and db, a string referring to the database alias the instance was loaded from or saved to.

Newly instantiated instances have adding=True and db=None, since they are yet to be saved. Instances fetched from a QuerySet will have adding=False and db set to the alias of the associated database.