django/docs/ref/models/relations.txt

=========================
Related objects reference
=========================

.. currentmodule:: django.db.models.fields.related

.. class:: RelatedManager

    A "related manager" is a manager used in a one-to-many or many-to-many
    related context. This happens in two cases:

    * The "other side" of a :class:`~django.db.models.ForeignKey` relation.
      That is::

            from django.db import models


            class Blog(models.Model):
                # ...
                pass


            class Entry(models.Model):
                blog = models.ForeignKey(Blog, on_delete=models.CASCADE, null=True)

      In the above example, the methods below will be available on
      the manager ``blog.entry_set``.

    * Both sides of a :class:`~django.db.models.ManyToManyField` relation
      ::

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


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

      In this example, the methods below will be available both on
      ``topping.pizza_set`` and on ``pizza.toppings``.

    .. method:: add(*objs, bulk=True, through_defaults=None)
    .. method:: aadd(*objs, bulk=True, through_defaults=None)

        *Asynchronous version*: ``aadd``

        Adds the specified model objects to the related object set.

        Example:

        .. code-block:: pycon

            >>> b = Blog.objects.get(id=1)
            >>> e = Entry.objects.get(id=234)
            >>> b.entry_set.add(e)  # Associates Entry e with Blog b.

        In the example above, in the case of a
        :class:`~django.db.models.ForeignKey` relationship,
        :meth:`QuerySet.update() <django.db.models.query.QuerySet.update>`
        is used to perform the update. This requires the objects to already be
        saved.

        You can use the ``bulk=False`` argument to instead have the related
        manager perform the update by calling ``e.save()``.

        Using ``add()`` with a many-to-many relationship, however, will not
        call any ``save()`` methods (the ``bulk`` argument doesn't exist), but
        rather create the relationships using :meth:`QuerySet.bulk_create()
        <django.db.models.query.QuerySet.bulk_create>`. If you need to execute
        some custom logic when a relationship is created, listen to the
        :data:`~django.db.models.signals.m2m_changed` signal, which will
        trigger ``pre_add`` and ``post_add`` actions.

        Using ``add()`` on a relation that already exists won't duplicate the
        relation, but it will still trigger signals.

        For many-to-many relationships ``add()`` accepts either model instances
        or field values, normally primary keys, as the ``*objs`` argument.

        Use the ``through_defaults`` argument to specify values for the new
        :ref:`intermediate model <intermediary-manytomany>` instance(s), if
        needed. You can use callables as values in the ``through_defaults``
        dictionary and they will be evaluated once before creating any
        intermediate instance(s).

        .. versionchanged:: 4.2

            ``aadd()`` method was added.

    .. method:: create(through_defaults=None, **kwargs)
    .. method:: acreate(through_defaults=None, **kwargs)

        *Asynchronous version*: ``acreate``

        Creates a new object, saves it and puts it in the related object set.
        Returns the newly created object:

        .. code-block:: pycon

            >>> b = Blog.objects.get(id=1)
            >>> e = b.entry_set.create(
            ...     headline="Hello", body_text="Hi", pub_date=datetime.date(2005, 1, 1)
            ... )

            # No need to call e.save() at this point -- it's already been saved.

        This is equivalent to (but simpler than):

        .. code-block:: pycon

            >>> b = Blog.objects.get(id=1)
            >>> e = Entry(blog=b, headline="Hello", body_text="Hi", pub_date=datetime.date(2005, 1, 1))
            >>> e.save(force_insert=True)

        Note that there's no need to specify the keyword argument of the model
        that defines the relationship. In the above example, we don't pass the
        parameter ``blog`` to ``create()``. Django figures out that the new
        ``Entry`` object's ``blog`` field should be set to ``b``.

        Use the ``through_defaults`` argument to specify values for the new
        :ref:`intermediate model <intermediary-manytomany>` instance, if
        needed. You can use callables as values in the ``through_defaults``
        dictionary.

        .. versionchanged:: 4.1

            ``acreate()`` method was added.

    .. method:: remove(*objs, bulk=True)
    .. method:: aremove(*objs, bulk=True)

        *Asynchronous version*: ``aremove``

        Removes the specified model objects from the related object set:

        .. code-block:: pycon

            >>> b = Blog.objects.get(id=1)
            >>> e = Entry.objects.get(id=234)
            >>> b.entry_set.remove(e)  # Disassociates Entry e from Blog b.

        Similar to :meth:`add()`, ``e.save()`` is called in the example above
        to perform the update. Using ``remove()`` with a many-to-many
        relationship, however, will delete the relationships using
        :meth:`QuerySet.delete()<django.db.models.query.QuerySet.delete>` which
        means no model ``save()`` methods are called; listen to the
        :data:`~django.db.models.signals.m2m_changed` signal if you wish to
        execute custom code when a relationship is deleted.

        For many-to-many relationships ``remove()`` accepts either model
        instances or field values, normally primary keys, as the ``*objs``
        argument.

        For :class:`~django.db.models.ForeignKey` objects, this method only
        exists if ``null=True``. If the related field can't be set to ``None``
        (``NULL``), then an object can't be removed from a relation without
        being added to another. In the above example, removing ``e`` from
        ``b.entry_set()`` is equivalent to doing ``e.blog = None``, and because
        the ``blog`` :class:`~django.db.models.ForeignKey` doesn't have
        ``null=True``, this is invalid.

        For :class:`~django.db.models.ForeignKey` objects, this method accepts
        a ``bulk`` argument to control how to perform the operation.
        If ``True`` (the default), ``QuerySet.update()`` is used.
        If ``bulk=False``, the ``save()`` method of each individual model
        instance is called instead. This triggers the
        :data:`~django.db.models.signals.pre_save` and
        :data:`~django.db.models.signals.post_save` signals and comes at the
        expense of performance.

        For many-to-many relationships, the ``bulk`` keyword argument doesn't
        exist.

        .. versionchanged:: 4.2

            ``aremove()`` method was added.

    .. method:: clear(bulk=True)
    .. method:: aclear(bulk=True)

        *Asynchronous version*: ``aclear``

        Removes all objects from the related object set:

        .. code-block:: pycon

            >>> b = Blog.objects.get(id=1)
            >>> b.entry_set.clear()

        Note this doesn't delete the related objects -- it just disassociates
        them.

        Just like ``remove()``, ``clear()`` is only available on
        :class:`~django.db.models.ForeignKey`\s where ``null=True`` and it also
        accepts the ``bulk`` keyword argument.

        For many-to-many relationships, the ``bulk`` keyword argument doesn't
        exist.

        .. versionchanged:: 4.2

            ``aclear()`` method was added.

    .. method:: set(objs, bulk=True, clear=False, through_defaults=None)
    .. method:: aset(objs, bulk=True, clear=False, through_defaults=None)

        *Asynchronous version*: ``aset``

        Replace the set of related objects:

        .. code-block:: pycon

            >>> new_list = [obj1, obj2, obj3]
            >>> e.related_set.set(new_list)

        This method accepts a ``clear`` argument to control how to perform the
        operation. If ``False`` (the default), the elements missing from the
        new set are removed using ``remove()`` and only the new ones are added.
        If ``clear=True``, the ``clear()`` method is called instead and the
        whole set is added at once.

        For :class:`~django.db.models.ForeignKey` objects, the ``bulk``
        argument is passed on to :meth:`add` and :meth:`remove`.

        For many-to-many relationships, the ``bulk`` keyword argument doesn't
        exist.

        Note that since ``set()`` is a compound operation, it is subject to
        race conditions. For instance, new objects may be added to the database
        in between the call to ``clear()`` and the call to ``add()``.

        For many-to-many relationships ``set()`` accepts a list of either model
        instances or field values, normally primary keys, as the ``objs``
        argument.

        Use the ``through_defaults`` argument to specify values for the new
        :ref:`intermediate model <intermediary-manytomany>` instance(s), if
        needed. You can use callables as values in the ``through_defaults``
        dictionary and they will be evaluated once before creating any
        intermediate instance(s).

        .. versionchanged:: 4.2

            ``aset()`` method was added.

    .. note::

       Note that ``add()``, ``aadd()``, ``create()``, ``acreate()``,
       ``remove()``, ``aremove()``, ``clear()``, ``aclear()``, ``set()``, and
       ``aset()`` all apply database changes immediately for all types of
       related fields. In other words, there is no need to call
       ``save()``/``asave()`` on either end of the relationship.

       If you use :meth:`~django.db.models.query.QuerySet.prefetch_related`,
       the ``add()``, ``aadd()``, ``remove()``, ``aremove()``, ``clear()``,
       ``aclear()``, ``set()``, and ``aset()`` methods clear the prefetched
       cache.