django/docs/topics/db/models.txt

.. _topics-db-models:

==============
Writing models
==============

.. module:: django.db.models

A model is the single, definitive source of data about your data. It contains
the essential fields and behaviors of the data you're storing. Generally, each
model maps to a single database table.

The basics:

    * Each model is a Python class that subclasses
      :class:`django.db.models.Model`.

    * Each attribute of the model represents a database field.

    * With all of this, Django gives you an automatically-generated
      database-access API; see :ref:`topics-db-queries`.

.. seealso::

    A companion to this document is the `official repository of model
    examples`_. (In the Django source distribution, these examples are in the
    ``tests/modeltests`` directory.)

    .. _official repository of model examples: http://www.djangoproject.com/documentation/models/

Quick example
=============

This example model defines a ``Person``, which has a ``first_name`` and
``last_name``::

    from django.db import models

    class Person(models.Model):
        first_name = models.CharField(max_length=30)
        last_name = models.CharField(max_length=30)

``first_name`` and ``last_name`` are fields_ of the model. Each field is
specified as a class attribute, and each attribute maps to a database column.

The above ``Person`` model would create a database table like this:

.. code-block:: sql

    CREATE TABLE myapp_person (
        "id" serial NOT NULL PRIMARY KEY,
        "first_name" varchar(30) NOT NULL,
        "last_name" varchar(30) NOT NULL
    );

Some technical notes:

    * The name of the table, ``myapp_person``, is automatically derived from
      some model metadata but can be overridden. See :ref:`table-names` for more
      details..

    * An ``id`` field is added automatically, but this behavior can be
      overridden. See :ref:`automatic-primary-key-fields`.

    * The ``CREATE TABLE`` SQL in this example is formatted using PostgreSQL
      syntax, but it's worth noting Django uses SQL tailored to the database
      backend specified in your :ref:`settings file <topics-settings>`.

Using models
============

Once you have defined your models, you need to tell Django you're going to *use*
those models. Do this by editing your settings file and changing the
:setting:`INSTALLED_APPS` setting to add the name of the module that contains
your ``models.py``.

For example, if the models for your application live in the module
``mysite.myapp.models`` (the package structure that is created for an
application by the :djadmin:`manage.py startapp <startapp>` script),
:setting:`INSTALLED_APPS` should read, in part::

    INSTALLED_APPS = (
        #...
        'mysite.myapp',
        #...
    )

When you add new apps to :setting:`INSTALLED_APPS`, be sure to run
:djadmin:`manage.py syncdb <syncdb>`.

Fields
======

The most important part of a model -- and the only required part of a model --
is the list of database fields it defines. Fields are specified by class
attributes.

Example::

    class Musician(models.Model):
        first_name = models.CharField(max_length=50)
        last_name = models.CharField(max_length=50)
        instrument = models.CharField(max_length=100)

    class Album(models.Model):
        artist = models.ForeignKey(Musician)
        name = models.CharField(max_length=100)
        release_date = models.DateField()
        num_stars = models.IntegerField()

Field types
-----------

Each field in your model should be an instance of the appropriate
:class:`~django.db.models.Field` class. Django uses the field class types to
determine a few things:

    * The database column type (e.g. ``INTEGER``, ``VARCHAR``).

    * The widget to use in Django's admin interface, if you care to use it
      (e.g. ``<input type="text">``, ``<select>``).

    * The minimal validation requirements, used in Django's admin and in
      automatically-generated forms.

Django ships with dozens of built-in field types; you can find the complete list
in the :ref:`model field reference <model-field-types>`. You can easily write
your own fields if Django's built-in ones don't do the trick; see
:ref:`howto-custom-model-fields`.

Field options
-------------

Each field takes a certain set of field-specific arguments (documented in the
:ref:`model field reference <model-field-types>`). For example,
:class:`~django.db.models.CharField` (and its subclasses) require a
:attr:`~django.db.models.CharField.max_length` argument which specifies the size
of the ``VARCHAR`` database field used to store the data.

There's also a set of common arguments available to all field types. All are
optional. They're fully explained in the :ref:`reference
<common-model-field-options>`, but here's a quick summary of the most often-used
ones:

    :attr:`~Field.null`
        If ``True``, Django will store empty values as ``NULL`` in the database.
        Default is ``False``.

    :attr:`~Field.blank`
        If ``True``, the field is allowed to be blank. Default is ``False``.

        Note that this is different than :attr:`~Field.null`.
        :attr:`~Field.null` is purely database-related, whereas
        :attr:`~Field.blank` is validation-related. If a field has
        :attr:`blank=True <Field.blank>`, validation on Django's admin site will
        allow entry of an empty value. If a field has :attr:`blank=False
        <Field.blank>`, the field will be required.

    :attr:`~Field.choices`
        An iterable (e.g., a list or tuple) of 2-tuples to use as choices for
        this field. If this is given, Django's admin will use a select box
        instead of the standard text field and will limit choices to the choices
        given.

        A choices list looks like this::

            YEAR_IN_SCHOOL_CHOICES = (
                ('FR', 'Freshman'),
                ('SO', 'Sophomore'),
                ('JR', 'Junior'),
                ('SR', 'Senior'),
                ('GR', 'Graduate'),
            )

    :attr:`~Field.default`
        The default value for the field. This can be a value or a callable
        object. If callable it will be called every time a new object is
        created.

    :attr:`~Field.help_text`
        Extra "help" text to be displayed under the field on the object's admin
        form. It's useful for documentation even if your object doesn't have an
        admin form.

    :attr:`~Field.primary_key`
        If ``True``, this field is the primary key for the model.

        If you don't specify :attr:`primary_key=True <Field.primary_key>` for
        any fields in your model, Django will automatically add an
        :class:`IntegerField` to hold the primary key, so you don't need to set
        :attr:`primary_key=True <Field.primary_key>` on any of your fields
        unless you want to override the default primary-key behavior. For more,
        see :ref:`automatic-primary-key-fields`.

    :attr:`~Field.unique`
        If ``True``, this field must be unique throughout the table.

Again, these are just short descriptions of the most common field options. Full
details can be found in the :ref:`common model field option reference
<common-model-field-options>`.

.. _automatic-primary-key-fields:

Automatic primary key fields
----------------------------

By default, Django gives each model the following field::

    id = models.AutoField(primary_key=True)

This is an auto-incrementing primary key.

If you'd like to specify a custom primary key, just specify
:attr:`primary_key=True <Field.primary_key>` on one of your fields. If Django
sees you've explicitly set :attr:`Field.primary_key`, it won't add the automatic
``id`` column.

Each model requires exactly one field to have :attr:`primary_key=True
<Field.primary_key>`.

.. _verbose-field-names:

Verbose field names
-------------------

Each field type, except for :class:`~django.db.models.ForeignKey`,
:class:`~django.db.models.ManyToManyField` and
:class:`~django.db.models.OneToOneField`, takes an optional first positional
argument -- a verbose name. If the verbose name isn't given, Django will
automatically create it using the field's attribute name, converting underscores
to spaces.

In this example, the verbose name is ``"Person's first name"``::

    first_name = models.CharField("Person's first name", max_length=30)

In this example, the verbose name is ``"first name"``::

    first_name = models.CharField(max_length=30)

:class:`~django.db.models.ForeignKey`,
:class:`~django.db.models.ManyToManyField` and
:class:`~django.db.models.OneToOneField` require the first argument to be a
model class, so use the :attr:`~Field.verbose_name` keyword argument::

    poll = models.ForeignKey(Poll, verbose_name="the related poll")
    sites = models.ManyToManyField(Site, verbose_name="list of sites")
    place = models.OneToOneField(Place, verbose_name="related place")

The convention is not to capitalize the first letter of the
:attr:`~Field.verbose_name`. Django will automatically capitalize the first
letter where it needs to.

Relationships
-------------

Clearly, the power of relational databases lies in relating tables to each
other. Django offers ways to define the three most common types of database
relationships: many-to-one, many-to-many and one-to-one.

Many-to-one relationships
~~~~~~~~~~~~~~~~~~~~~~~~~

To define a many-to-one relationship, use :class:`~django.db.models.ForeignKey`.
You use it just like any other :class:`~django.db.models.Field` type: by
including it as a class attribute of your model.

:class:`~django.db.models.ForeignKey` requires a positional argument: the class
to which the model is related.

For example, if a ``Car`` model has a ``Manufacturer`` -- that is, a
``Manufacturer`` makes multiple cars but each ``Car`` only has one
``Manufacturer`` -- use the following definitions::

    class Manufacturer(models.Model):
        # ...

    class Car(models.Model):
        manufacturer = models.ForeignKey(Manufacturer)
        # ...

You can also create :ref:`recursive relationships <recursive-relationships>` (an
object with a many-to-one relationship to itself) and :ref:`relationships to
models not yet defined <lazy-relationships>`; see :ref:`the model field
reference <ref-foreignkey>` for details.

It's suggested, but not required, that the name of a
:class:`~django.db.models.ForeignKey` field (``manufacturer`` in the example
above) be the name of the model, lowercase. You can, of course, call the field
whatever you want. For example::

    class Car(models.Model):
        company_that_makes_it = models.ForeignKey(Manufacturer)
        # ...

.. seealso::

    See the `Many-to-one relationship model example`_ for a full example.

.. _Many-to-one relationship model example: http://www.djangoproject.com/documentation/models/many_to_one/

:class:`~django.db.models.ForeignKey` fields also accept a number of extra
arguments which are explained in :ref:`the model field reference
<foreign-key-arguments>`. These options help define how the relationship should
work; all are optional.

Many-to-many relationships
~~~~~~~~~~~~~~~~~~~~~~~~~~

To define a many-to-many relationship, use
:class:`~django.db.models.ManyToManyField`. You use it just like any other
:class:`~django.db.models.Field` type: by including it as a class attribute of
your model.

:class:`~django.db.models.ManyToManyField` requires a positional argument: the
class to which the model is related.

For example, if a ``Pizza`` has multiple ``Topping`` objects -- that is, a
``Topping`` can be on multiple pizzas and each ``Pizza`` has multiple toppings
-- here's how you'd represent that::

    class Topping(models.Model):
        # ...

    class Pizza(models.Model):
        # ...
        toppings = models.ManyToManyField(Topping)

As with :class:`~django.db.models.ForeignKey`, you can also create
:ref:`recursive relationships <recursive-relationships>` (an object with a
many-to-one relationship to itself) and :ref:`relationships to models not yet
defined <lazy-relationships>`; see :ref:`the model field reference
<ref-manytomany>` for details.

It's suggested, but not required, that the name of a
:class:`~django.db.models.ManyToManyField` (``toppings`` in the example above)
be a plural describing the set of related model objects.

It doesn't matter which model gets the
:class:`~django.db.models.ManyToManyField`, but you only need it in one of the
models -- not in both.

Generally, :class:`~django.db.models.ManyToManyField` instances should go in the
object that's going to be edited in the admin interface, if you're using
Django's admin. In the above example, ``toppings`` is in ``Pizza`` (rather than
``Topping`` having a ``pizzas`` :class:`~django.db.models.ManyToManyField` )
because it's more natural to think about a pizza having toppings than a
topping being on multiple pizzas. The way it's set up above, the ``Pizza`` admin
form would let users select the toppings.

.. seealso::

    See the `Many-to-many relationship model example`_ for a full example.

.. _Many-to-many relationship model example: http://www.djangoproject.com/documentation/models/many_to_many/

:class:`~django.db.models.ManyToManyField` fields also accept a number of extra
arguments which are explained in :ref:`the model field reference
<manytomany-arguments>`. These options help define how the relationship should
work; all are optional.

.. _intermediary-manytomany:

Extra fields on many-to-many relationships
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

.. versionadded:: 1.0

When you're only dealing with simple many-to-many relationships such as
mixing and matching pizzas and toppings, a standard :class:`~django.db.models.ManyToManyField` is all you need. However, sometimes
you may need to associate data with the relationship between two models.

For example, consider the case of an application tracking the musical groups
which musicians belong to. There is a many-to-many relationship between a person
and the groups of which they are a member, so you could use a
:class:`~django.db.models.ManyToManyField` to represent this relationship.
However, there is a lot of detail about the membership that you might want to
collect, such as the date at which the person joined the group.

For these situations, Django allows you to specify the model that will be used
to govern the many-to-many relationship. You can then put extra fields on the
intermediate model. The intermediate model is associated with the
:class:`~django.db.models.ManyToManyField` using the
:attr:`through <ManyToManyField.through>` argument to point to the model
that will act as an intermediary. For our musician example, the code would look
something like this::

    class Person(models.Model):
        name = models.CharField(max_length=128)

        def __unicode__(self):
            return self.name

    class Group(models.Model):
        name = models.CharField(max_length=128)
        members = models.ManyToManyField(Person, through='Membership')

        def __unicode__(self):
            return self.name

    class Membership(models.Model):
        person = models.ForeignKey(Person)
        group = models.ForeignKey(Group)
        date_joined = models.DateField()
        invite_reason = models.CharField(max_length=64)

When you set up the intermediary model, you explicitly specify foreign
keys to the models that are involved in the ManyToMany relation. This
explicit declaration defines how the two models are related.

There are a few restrictions on the intermediate model:

    * Your intermediate model must contain one - and *only* one - foreign key
      to the target model (this would be ``Person`` in our example). If you
      have more than one foreign key, a validation error will be raised.

    * Your intermediate model must contain one - and *only* one - foreign key
      to the source model (this would be ``Group`` in our example). If you
      have more than one foreign key, a validation error will be raised.

    * The only exception to this is a model which has a many-to-many
      relationship to itself, through an intermediary model. In this
      case, two foreign keys to the same model are permitted, but they
      will be treated as the two (different) sides of the many-to-many
      relation.

    * When defining a many-to-many relationship from a model to
      itself, using an intermediary model, you *must* use
      :attr:`symmetrical=False <ManyToManyField.symmetrical>` (see
      :ref:`the model field reference <manytomany-arguments>`).

Now that you have set up your :class:`~django.db.models.ManyToManyField` to use
your intermediary model (``Membership``, in this case), you're ready to start
creating some many-to-many relationships. You do this by creating instances of
the intermediate model::

    >>> ringo = Person.objects.create(name="Ringo Starr")
    >>> paul = Person.objects.create(name="Paul McCartney")
    >>> beatles = Group.objects.create(name="The Beatles")
    >>> m1 = Membership(person=ringo, group=beatles,
    ...     date_joined=date(1962, 8, 16),
    ...     invite_reason= "Needed a new drummer.")
    >>> m1.save()
    >>> beatles.members.all()
    [<Person: Ringo Starr>]
    >>> ringo.group_set.all()
    [<Group: The Beatles>]
    >>> m2 = Membership.objects.create(person=paul, group=beatles,
    ...     date_joined=date(1960, 8, 1),
    ...     invite_reason= "Wanted to form a band.")
    >>> beatles.members.all()
    [<Person: Ringo Starr>, <Person: Paul McCartney>]

Unlike normal many-to-many fields, you *can't* use ``add``, ``create``,
or assignment (i.e., ``beatles.members = [...]``) to create relationships::

    # THIS WILL NOT WORK
    >>> beatles.members.add(john)
    # NEITHER WILL THIS
    >>> beatles.members.create(name="George Harrison")
    # AND NEITHER WILL THIS
    >>> beatles.members = [john, paul, ringo, george]

Why? You can't just create a relationship between a ``Person`` and a ``Group``
- you need to specify all the detail for the relationship required by the
``Membership`` model. The simple ``add``, ``create`` and assignment calls
don't provide a way to specify this extra detail. As a result, they are
disabled for many-to-many relationships that use an intermediate model.
The only way to create this type of relationship is to create instances of the
intermediate model.

The ``remove`` method is disabled for similar reasons. However, the
``clear()`` method can be used to remove all many-to-many relationships
for an instance::

    # Beatles have broken up
    >>> beatles.members.clear()

Once you have established the many-to-many relationships by creating instances
of your intermediate model, you can issue queries. Just as with normal
many-to-many relationships, you can query using the attributes of the
many-to-many-related model::

    # Find all the groups with a member whose name starts with 'Paul'
    >>> Groups.objects.filter(members__name__startswith='Paul')
    [<Group: The Beatles>]

As you are using an intermediate model, you can also query on its attributes::

    # Find all the members of the Beatles that joined after 1 Jan 1961
    >>> Person.objects.filter(
    ...     group__name='The Beatles',
    ...     membership__date_joined__gt=date(1961,1,1))
    [<Person: Ringo Starr]


One-to-one relationships
~~~~~~~~~~~~~~~~~~~~~~~~

To define a one-to-one relationship, use
:class:`~django.db.models.OneToOneField`. You use it just like any other
``Field`` type: by including it as a class attribute of your model.

This is most useful on the primary key of an object when that object "extends"
another object in some way.

:class:`~django.db.models.OneToOneField` requires a positional argument: the
class to which the model is related.

For example, if you were building a database of "places", you would
build pretty standard stuff such as address, phone number, etc. in the
database. Then, if you wanted to build a database of restaurants on
top of the places, instead of repeating yourself and replicating those
fields in the ``Restaurant`` model, you could make ``Restaurant`` have
a :class:`~django.db.models.OneToOneField` to ``Place`` (because a
restaurant "is a" place; in fact, to handle this you'd typically use
:ref:`inheritance <model-inheritance>`, which involves an implicit
one-to-one relation).

As with :class:`~django.db.models.ForeignKey`, a
:ref:`recursive relationship <recursive-relationships>`
can be defined and
:ref:`references to as-yet undefined models <lazy-relationships>`
can be made; see :ref:`the model field reference <ref-onetoone>` for details.

.. seealso::

    See the `One-to-one relationship model example`_ for a full example.

.. _One-to-one relationship model example: http://www.djangoproject.com/documentation/models/one_to_one/

.. versionadded:: 1.0

:class:`~django.db.models.OneToOneField` fields also accept one optional argument
described in the :ref:`model field reference <ref-onetoone>`.

:class:`~django.db.models.OneToOneField` classes used to automatically become
the primary key on a model. This is no longer true (although you can manually
pass in the :attr:`~django.db.models.Field.primary_key` argument if you like).
Thus, it's now possible to have multiple fields of type
:class:`~django.db.models.OneToOneField` on a single model.

Models across files
-------------------

It's perfectly OK to relate a model to one from another app. To do this,
import the related model at the top of the model that holds your model. Then,
just refer to the other model class wherever needed. For example::

    from mysite.geography.models import ZipCode

    class Restaurant(models.Model):
        # ...
        zip_code = models.ForeignKey(ZipCode)

Field name restrictions
-----------------------

Django places only two restrictions on model field names:

    1. A field name cannot be a Python reserved word, because that would result
       in a Python syntax error. For example::

           class Example(models.Model):
               pass = models.IntegerField() # 'pass' is a reserved word!

    2. A field name cannot contain more than one underscore in a row, due to
       the way Django's query lookup syntax works. For example::

           class Example(models.Model):
               foo__bar = models.IntegerField() # 'foo__bar' has two underscores!

These limitations can be worked around, though, because your field name doesn't
necessarily have to match your database column name. See the
:attr:`~Field.db_column` option.

SQL reserved words, such as ``join``, ``where`` or ``select``, *are* allowed as
model field names, because Django escapes all database table names and column
names in every underlying SQL query. It uses the quoting syntax of your
particular database engine.

Custom field types
------------------

.. versionadded:: 1.0

If one of the existing model fields cannot be used to fit your purposes, or if
you wish to take advantage of some less common database column types, you can
create your own field class. Full coverage of creating your own fields is
provided in :ref:`howto-custom-model-fields`.

.. _meta-options:

Meta options
============

Give your model metadata by using an inner ``class Meta``, like so::

    class Ox(models.Model):
        horn_length = models.IntegerField()

        class Meta:
            ordering = ["horn_length"]
            verbose_name_plural = "oxen"

Model metadata is "anything that's not a field", such as ordering options
(:attr:`~Options.ordering`), database table name (:attr:`~Options.db_table`), or
human-readable singular and plural names (:attr:`~Options.verbose_name` and
:attr:`~Options.verbose_name_plural`). None are required, and adding ``class
Meta`` to a model is completely optional.

A complete list of all possible ``Meta`` options can be found in the :ref:`model
option reference <ref-models-options>`.

.. _model-methods:

Model methods
=============

Define custom methods on a model to add custom "row-level" functionality to your
objects. Whereas :class:`~django.db.models.Manager` methods are intended to do
"table-wide" things, model methods should act on a particular model instance.

This is a valuable technique for keeping business logic in one place -- the
model.

For example, this model has a few custom methods::

    from django.contrib.localflavor.us.models import USStateField

    class Person(models.Model):
        first_name = models.CharField(max_length=50)
        last_name = models.CharField(max_length=50)
        birth_date = models.DateField()
        address = models.CharField(max_length=100)
        city = models.CharField(max_length=50)
        state = USStateField() # Yes, this is America-centric...

        def baby_boomer_status(self):
            "Returns the person's baby-boomer status."
            import datetime
            if datetime.date(1945, 8, 1) <= self.birth_date <= datetime.date(1964, 12, 31):
                return "Baby boomer"
            if self.birth_date < datetime.date(1945, 8, 1):
                return "Pre-boomer"
            return "Post-boomer"

        def is_midwestern(self):
            "Returns True if this person is from the Midwest."
            return self.state in ('IL', 'WI', 'MI', 'IN', 'OH', 'IA', 'MO')

        def _get_full_name(self):
            "Returns the person's full name."
            return '%s %s' % (self.first_name, self.last_name)
        full_name = property(_get_full_name)

The last method in this example is a :term:`property`. `Read more about
properties`_.

.. _Read more about properties: http://www.python.org/download/releases/2.2/descrintro/#property

The :ref:`model instance reference <ref-models-instances>` has a complete list
of :ref:`methods automatically given to each model <model-instance-methods>`.
You can override most of these -- see `overriding predefined model methods`_,
below -- but there are a couple that you'll almost always want to define:

    :meth:`~Model.__unicode__`
        A Python "magic method" that returns a unicode "representation" of any
        object. This is what Python and Django will use whenever a model
        instance needs to be coerced and displayed as a plain string. Most
        notably, this happens when you display an object in an interactive
        console or in the admin.

        You'll always want to define this method; the default isn't very helpful
        at all.

    :meth:`~Model.get_absolute_url`
        This tells Django how to calculate the URL for an object. Django uses
        this in its admin interface, and any time it needs to figure out a URL
        for an object.

        Any object that has a URL that uniquely identifies it should define this
        method.

Overriding predefined model methods
-----------------------------------

There's another set of :ref:`model methods <model-instance-methods>` that
encapsulate a bunch of database behavior that you'll want to customize. In
particular you'll often want to change the way :meth:`~Model.save` and
:meth:`~Model.delete` work.

You're free to override these methods (and any other model method) to alter
behavior.

A classic use-case for overriding the built-in methods is if you want something
to happen whenever you save an object. For example (see
:meth:`~Model.save` for documentation of the parameters it accepts)::

    class Blog(models.Model):
        name = models.CharField(max_length=100)
        tagline = models.TextField()

        def save(self, force_insert=False, force_update=False):
            do_something()
            super(Blog, self).save(force_insert, force_update) # Call the "real" save() method.
            do_something_else()

You can also prevent saving::

    class Blog(models.Model):
        name = models.CharField(max_length=100)
        tagline = models.TextField()

        def save(self, force_insert=False, force_update=False):
            if self.name == "Yoko Ono's blog":
                return # Yoko shall never have her own blog!
            else:
                super(Blog, self).save(force_insert, force_update) # Call the "real" save() method.

It's important to remember to call the superclass method -- that's that
``super(Blog, self).save()`` business -- to ensure that the object still gets
saved into the database. If you forget to call the superclass method, the
default behavior won't happen and the database won't get touched.

Executing custom SQL
--------------------

Another common pattern is writing custom SQL statements in model methods and
module-level methods. The object :class:`django.db.connection
<django.db.backends.DatabaseWrapper>` represents the current database
connection. To use it, call :meth:`connection.cursor()
<django.db.backends.DatabaseWrapper.cursor>` to get a cursor object. Then, call
``cursor.execute(sql, [params])`` to execute the SQL and
:meth:`cursor.fetchone() <django.db.backends.CursorWrapper.fetchone>` or
:meth:`cursor.fetchall() <django.db.backends.CursorWrapper.fetchall>` to return
the resulting rows. For example::

    def my_custom_sql(self):
        from django.db import connection
        cursor = connection.cursor()
        cursor.execute("SELECT foo FROM bar WHERE baz = %s", [self.baz])
        row = cursor.fetchone()
        return row

:class:`connection <django.db.backends.DatabaseWrapper>` and :class:`cursor
<django.db.backends.CursorWrapper>` mostly implement the standard Python
DB-API -- see :pep:`249` -- with the addition of Django's :ref:`transaction
handling <topics-db-transactions>`. If you're not familiar with the Python
DB-API, note that the SQL statement in :meth:`cursor.execute()
<django.db.backends.CursorWrapper.execute>` uses placeholders, ``"%s"``, rather
than adding parameters directly within the SQL. If you use this technique, the
underlying database library will automatically add quotes and escaping to your
parameter(s) as necessary. (Also note that Django expects the ``"%s"``
placeholder, *not* the ``"?"`` placeholder, which is used by the SQLite Python
bindings. This is for the sake of consistency and sanity.)

A final note: If all you want to do is a custom ``WHERE`` clause, you can use
the :meth:`~QuerySet.extra` lookup method, which lets you add custom SQL to a
query.

.. _model-inheritance:

Model inheritance
=================

.. versionadded:: 1.0

Model inheritance in Django works almost identically to the way normal
class inheritance works in Python. The only decision you have to make
is whether you want the parent models to be models in their own right
(with their own database tables), or if the parents are just holders
of common information that will only be visible through the child
models.

There are three styles of inheritance that are possible in Django.

 1. Often, you will just want to use the parent class to hold information that
    you don't want to have to type out for each child model. This class isn't
    going to ever be used in isolation, so :ref:`abstract-base-classes` are
    what you're after.
 2. If you're subclassing an existing model (perhaps something from another
    application entirely) and want each model to have its own database table,
    :ref:`multi-table-inheritance` is the way to go.
 3. Finally, if you only want to modify the Python-level behaviour of a model,
    without changing the models fields in any way, you can use
    :ref:`proxy-models`.

.. _abstract-base-classes:

Abstract base classes
---------------------

Abstract base classes are useful when you want to put some common
information into a number of other models. You write your base class
and put ``abstract=True`` in the :ref:`Meta <meta-options>`
class. This model will then not be used to create any database
table. Instead, when it is used as a base class for other models, its
fields will be added to those of the child class. It is an error to
have fields in the abstract base class with the same name as those in
the child (and Django will raise an exception).

An example::

    class CommonInfo(models.Model):
        name = models.CharField(max_length=100)
        age = models.PositiveIntegerField()

        class Meta:
            abstract = True

    class Student(CommonInfo):
        home_group = models.CharField(max_length=5)

The ``Student`` model will have three fields: ``name``, ``age`` and
``home_group``. The ``CommonInfo`` model cannot be used as a normal Django
model, since it is an abstract base class. It does not generate a database
table or have a manager, and cannot be instantiated or saved directly.

For many uses, this type of model inheritance will be exactly what you want.
It provides a way to factor out common information at the Python level, whilst
still only creating one database table per child model at the database level.

``Meta`` inheritance
~~~~~~~~~~~~~~~~~~~~

When an abstract base class is created, Django makes any :ref:`Meta <meta-options>`
inner class you declared in the base class available as an
attribute. If a child class does not declare its own :ref:`Meta <meta-options>`
class, it will inherit the parent's :ref:`Meta <meta-options>`. If the child wants to
extend the parent's :ref:`Meta <meta-options>` class, it can subclass it. For example::

    class CommonInfo(models.Model):
        ...
        class Meta:
            abstract = True
            ordering = ['name']

    class Student(CommonInfo):
        ...
        class Meta(CommonInfo.Meta):
            db_table = 'student_info'

Django does make one adjustment to the :ref:`Meta <meta-options>` class of an abstract base
class: before installing the :ref:`Meta <meta-options>` attribute, it sets ``abstract=False``.
This means that children of abstract base classes don't automatically become
abstract classes themselves. Of course, you can make an abstract base class
that inherits from another abstract base class. You just need to remember to
explicitly set ``abstract=True`` each time.

Some attributes won't make sense to include in the :ref:`Meta <meta-options>` class of an
abstract base class. For example, including ``db_table`` would mean that all
the child classes (the ones that don't specify their own :ref:`Meta <meta-options>`) would use
the same database table, which is almost certainly not what you want.

.. _abstract-related-name:

Be careful with ``related_name``
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

If you are using the :attr:`~django.db.models.ForeignKey.related_name` attribute on a ``ForeignKey`` or
``ManyToManyField``, you must always specify a *unique* reverse name for the
field. This would normally cause a problem in abstract base classes, since the
fields on this class are included into each of the child classes, with exactly
the same values for the attributes (including :attr:`~django.db.models.ForeignKey.related_name`) each time.

To work around this problem, when you are using :attr:`~django.db.models.ForeignKey.related_name` in an
abstract base class (only), part of the name should be the string
``'%(class)s'``. This is replaced by the lower-cased name of the child class
that the field is used in. Since each class has a different name, each related
name will end up being different. For example::

    class Base(models.Model):
        m2m = models.ManyToManyField(OtherModel, related_name="%(class)s_related")

        class Meta:
            abstract = True

    class ChildA(Base):
        pass

    class ChildB(Base):
        pass

The reverse name of the ``ChildA.m2m`` field will be ``childa_related``,
whilst the reverse name of the ``ChildB.m2m`` field will be
``childb_related``. It is up to you how you use the ``'%(class)s'`` portion to
construct your related name, but if you forget to use it, Django will raise
errors when you validate your models (or run :djadmin:`syncdb`).

If you don't specify a :attr:`~django.db.models.ForeignKey.related_name` attribute for a field in an
abstract base class, the default reverse name will be the name of the
child class followed by ``'_set'``, just as it normally would be if
you'd declared the field directly on the child class. For example, in
the above code, if the :attr:`~django.db.models.ForeignKey.related_name` attribute was omitted, the
reverse name for the ``m2m`` field would be ``childa_set`` in the
``ChildA`` case and ``childb_set`` for the ``ChildB`` field.

.. _multi-table-inheritance:

Multi-table inheritance
-----------------------

The second type of model inheritance supported by Django is when each model in
the hierarchy is a model all by itself. Each model corresponds to its own
database table and can be queried and created individually. The inheritance
relationship introduces links between the child model and each of its parents
(via an automatically-created :class:`~django.db.models.fields.OneToOneField`).
For example::

    class Place(models.Model):
        name = models.CharField(max_length=50)
        address = models.CharField(max_length=80)

    class Restaurant(Place):
        serves_hot_dogs = models.BooleanField()
        serves_pizza = models.BooleanField()

All of the fields of ``Place`` will also be available in ``Restaurant``,
although the data will reside in a different database table. So these are both
possible::

    >>> Place.objects.filter(name="Bob's Cafe")
    >>> Restaurant.objects.filter(name="Bob's Cafe")

If you have a ``Place`` that is also a ``Restaurant``, you can get from the
``Place`` object to the ``Restaurant`` object by using the lower-case version
of the model name::

    >>> p = Place.objects.filter(name="Bob's Cafe")
    # If Bob's Cafe is a Restaurant object, this will give the child class:
    >>> p.restaurant
    <Restaurant: ...>

However, if ``p`` in the above example was *not* a ``Restaurant`` (it had been
created directly as a ``Place`` object or was the parent of some other class),
referring to ``p.restaurant`` would raise a Restaurant.DoesNotExist exception.

``Meta`` and multi-table inheritance
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

In the multi-table inheritance situation, it doesn't make sense for a child
class to inherit from its parent's :ref:`Meta <meta-options>` class. All the :ref:`Meta <meta-options>` options
have already been applied to the parent class and applying them again would
normally only lead to contradictory behavior (this is in contrast with the
abstract base class case, where the base class doesn't exist in its own
right).

So a child model does not have access to its parent's :ref:`Meta
<meta-options>` class. However, there are a few limited cases where the child
inherits behavior from the parent: if the child does not specify an
:attr:`django.db.models.Options.ordering` attribute or a
:attr:`django.db.models.Options.get_latest_by` attribute, it will inherit
these from its parent.

If the parent has an ordering and you don't want the child to have any natural
ordering, you can explicitly disable it::

    class ChildModel(ParentModel):
        ...
        class Meta:
            # Remove parent's ordering effect
            ordering = []

Inheritance and reverse relations
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Because multi-table inheritance uses an implicit
:class:`~django.db.models.fields.OneToOneField` to link the child and
the parent, it's possible to move from the parent down to the child,
as in the above example. However, this uses up the name that is the
default :attr:`~django.db.models.ForeignKey.related_name` value for
:class:`django.db.models.fields.ForeignKey` and
:class:`django.db.models.fields.ManyToManyField` relations.  If you
are putting those types of relations on a subclass of another model,
you **must** specify the
:attr:`~django.db.models.ForeignKey.related_name` attribute on each
such field. If you forget, Django will raise an error when you run
:djadmin:`validate` or :djadmin:`syncdb`.

For example, using the above ``Place`` class again, let's create another
subclass with a :class:`~django.db.models.fields.ManyToManyField`::

    class Supplier(Place):
        # Must specify related_name on all relations.
        customers = models.ManyToManyField(Restaurant, related_name='provider')


Specifying the parent link field
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

As mentioned, Django will automatically create a
:class:`~django.db.models.fields.OneToOneField` linking your child
class back any non-abstract parent models. If you want to control the
name of the attribute linking back to the parent, you can create your
own :class:`~django.db.models.fields.OneToOneField` and set
:attr:`parent_link=True <django.db.models.fields.OneToOneField.parent_link>`
to indicate that your field is the link back to the parent class.

.. _proxy-models:

Proxy models
------------

.. versionadded:: 1.1

When using :ref:`multi-table inheritance <multi-table-inheritance>`, a new
database table is created for each subclass of a model. This is usually the
desired behavior, since the subclass needs a place to store any additional
data fields that are not present on the base class. Sometimes, however, you
only want to change the Python behavior of a model -- perhaps to change the
default manager, or add a new method.

This is what proxy model inheritance is for: creating a *proxy* for the
original model. You can create, delete and update instances of the proxy model
and all the data will be saved as if you were using the original (non-proxied)
model. The difference is that you can change things like the default model
ordering or the default manager in the proxy, without having to alter the
original.

Proxy models are declared like normal models. You tell Django that it's a
proxy model by setting the :attr:`~django.db.models.Options.proxy` attribute of
the ``Meta`` class to ``True``.

For example, suppose you want to add a method to the standard ``User`` model
that will be used in your templates. You can do it like this::

    from django.contrib.auth.models import User

    class MyUser(User):
        class Meta:
            proxy = True

        def do_something(self):
            ...

The ``MyUser`` class operates on the same database table as its parent
``User`` class. In particular, any new instances of ``User`` will also be
accessible through ``MyUser``, and vice-versa::

    >>> u = User.objects.create(username="foobar")
    >>> MyUser.objects.get(username="foobar")
    <MyUser: foobar>

You could also use a proxy model to define a different default ordering on a
model. The standard ``User`` model has no ordering defined on it
(intentionally; sorting is expensive and we don't want to do it all the time
when we fetch users). You might want to regularly order by the ``username``
attribute when you use the proxy. This is easy::

    class OrderedUser(User):
        class Meta:
            ordering = ["username"]
            proxy = True

Now normal ``User`` queries will be unorderd and ``OrderedUser`` queries will
be ordered by ``username``.

Querysets still return the model that was requested
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

There is no way to have Django return, say, a ``MyUser`` object whenever you
query for ``User`` objects. A queryset for ``User`` objects will return those
types of objects. The whole point of proxy objects is that code relying on the
original ``User`` will use those and your own code can use the extensions you
included (that no other code is relying on anyway). It is not a way to replace
the ``User`` (or any other) model everywhere with something of your own
creation.

Base class restrictions
~~~~~~~~~~~~~~~~~~~~~~~

A proxy model must inherit from exactly one non-abstract model class. You
can't inherit from multiple non-abstract models as the proxy model doesn't
provide any connection between the rows in the different database tables. A
proxy model can inherit from any number of abstract model classes, providing
they do *not* define any model fields.

Proxy models inherit any ``Meta`` options that they don't define from their
non-abstract model parent (the model they are proxying for).

Proxy model managers
~~~~~~~~~~~~~~~~~~~~

If you don't specify any model managers on a proxy model, it inherits the
managers from its model parents. If you define a manager on the proxy model,
it will become the default, although any managers defined on the parent
classes will still be available.

Continuing our example from above, you could change the default manager used
when you query the ``User`` model like this::

    class NewManager(models.Manager):
        ...

    class MyUser(User):
        objects = NewManager()

        class Meta:
            proxy = True

If you wanted to add a new manager to the Proxy, without replacing the
existing default, you can use the techniques described in the :ref:`custom
manager <custom-managers-and-inheritance>` documentation: create a base class
containing the new managers and inherit that after the primary base class::

    # Create an abstract class for the new manager.
    class ExtraManagers:
        secondary = NewManager()

        class Meta:
            abstract = True

    class MyUser(User, ExtraManagers):
        class Meta:
            proxy = True

You probably won't need to do this very often, but, when you do, it's
possible.

.. _proxy-vs-unmanaged-models:

Differences between proxy inheritance and  unmanaged models
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Proxy model inheritance might look fairly similar to creating an unmanaged
model, using the :attr:`~django.db.models.Options.managed` attribute on a
model's ``Meta`` class. The two alternatives are not quite the same and it's
worth considering which one you should use.

One difference is that you can (and, in fact, must unless you want an empty
model) specify model fields on models with ``Meta.managed=False``. You could,
with careful setting of :attr:`Meta.db_table
<django.db.models.Options.db_table>` create an unmanaged model that shadowed
an existing model and add Python methods to it. However, that would be very
repetitive and fragile as you need to keep both copies synchronized if you
make any changes.

The other difference that is more important for proxy models, is how model
managers are handled. Proxy models are intended to behave exactly like the
model they are proxying for. So they inherit the parent model's managers,
including the default manager. In the normal multi-table model inheritance
case, children do not inherit managers from their parents as the custom
managers aren't always appropriate when extra fields are involved. The
:ref:`manager documentation <custom-managers-and-inheritance>` has more
details about this latter case.

When these two features were implemented, attempts were made to squash them
into a single option. It turned out that interactions with inheritance, in
general, and managers, in particular, made the API very complicated and
potentially difficult to understand and use. It turned out that two options
were needed in any case, so the current separation arose.

So, the general rules are:

    1. If you are mirroring an existing model or database table and don't want
       all the original database table columns, use ``Meta.managed=False``.
       That option is normally useful for modeling database views and tables
       not under the control of Django.
    2. If you are wanting to change the Python-only behavior of a model, but
       keep all the same fields as in the original, use ``Meta.proxy=True``.
       This sets things up so that the proxy model is an exact copy of the
       storage structure of the original model when data is saved.

Multiple inheritance
--------------------

Just as with Python's subclassing, it's possible for a Django model to inherit
from multiple parent models. Keep in mind that normal Python name resolution
rules apply. The first base class that a particular name (e.g. :ref:`Meta
<meta-options>`) appears in will be the one that is used; for example, this
means that if multiple parents contain a :ref:`Meta <meta-options>` class,
only the first one is going to be used, and all others will be ignored.

Generally, you won't need to inherit from multiple parents. The main use-case
where this is useful is for "mix-in" classes: adding a particular extra
field or method to every class that inherits the mix-in. Try to keep your
inheritance hierarchies as simple and straightforward as possible so that you
won't have to struggle to work out where a particular piece of information is
coming from.

Field name "hiding" is not permitted
-------------------------------------

In normal Python class inheritance, it is permissible for a child class to
override any attribute from the parent class. In Django, this is not permitted
for attributes that are :class:`~django.db.models.fields.Field` instances (at
least, not at the moment). If a base class has a field called ``author``, you
cannot create another model field called ``author`` in any class that inherits
from that base class.

Overriding fields in a parent model leads to difficulties in areas such as
initialising new instances (specifying which field is being intialised in
``Model.__init__``) and serialization. These are features which normal Python
class inheritance doesn't have to deal with in quite the same way, so the
difference between Django model inheritance and Python class inheritance isn't
merely arbitrary.

This restriction only applies to attributes which are
:class:`~django.db.models.fields.Field` instances. Normal Python attributes
can be overridden if you wish. It also only applies to the name of the
attribute as Python sees it: if you are manually specifying the database
column name, you can have the same column name appearing in both a child and
an ancestor model for multi-table inheritance (they are columns in two
different database tables).

Django will raise a ``FieldError`` exception if you override any model field
in any ancestor model.