django-computedfields provides autoupdated database fields for model methods.
Tested with Django 4.2 and 5.2 (Python 3.8 to 3.13).
Just derive your model from ComputedFieldsModel and place
the @computed decorator at a method:
from django.db import models
from computedfields.models import ComputedFieldsModel, computed
class MyModel(ComputedFieldsModel):
name = models.CharField(max_length=32)
@computed(models.CharField(max_length=32), depends=[('self', ['name'])])
def computed_field(self):
return self.name.upper()computed_field will be turned into a real database field
and can be accessed and searched like any other database field.
During saving the associated method gets called and it’s result
written to the database.
If you need to recalculate the computed field but without saving it, use
from computedfields.models import compute
>>> from computedfields.models import compute
>>> person = MyModel.objects.get(id=1) # this is to retrieve existing record
>>> person.computed_field # outputs 'BERTY'
>>> person.name = 'nina' # changing the dependent field `name` to nina
>>> compute(person, 'computed_field') # outputs 'NINA'
>>> person.computed_field # outputs 'BERTY' because the `person` is not yet saved
>>> person.save() # alters the database record for `name` and `computed_field`
>>> person.computed_field # outputs 'NINA'The depends keyword argument can be used with any relation to indicate dependencies to fields on other models as well:
from django.db import models
from computedfields.models import ComputedFieldsModel, computed
class MyModel(ComputedFieldsModel):
name = models.CharField(max_length=32)
fk = models.ForeignKey(SomeModel)
@computed(
models.CharField(max_length=32),
depends=[
('self', ['name']),
('fk', ['fieldname'])
]
)
def computed_field(self):
return self.name.upper() + self.fk.fieldnameNow changes to self.name, fk or fk.fieldname will update computed_field.
Instead of using the @computed decorator with inline field definitions,
you can also use a more declarative syntax with ComputedField, example from above rewritten:
from django.db import models
from computedfields.models import ComputedFieldsModel, ComputedField
def get_upper_string(inst):
return inst.name.upper()
class MyModel(ComputedFieldsModel):
name = models.CharField(max_length=32)
computed_field = ComputedField(
models.CharField(max_length=32),
depends=[('self', ['name'])],
compute=get_upper_string
)The documentation can be found here.
-
0.3.5
- performance improvement: allow select_related/prefetch_related on UNIONed resolver updates
-
0.3.4
- Fix UNIONed resolver updates, that contain select_related
-
0.3.3
- auto recovery for not_computed context
-
0.3.2
- regression fix (#190): add implicit fk dependencies from reverse relations
-
0.3.1 (discouraged due to regression #190)
- improved CF updates on through models
- limitations section in docs
-
0.3.0 - new beta release (discouraged due to regression #190)
- new features:
- default_on_create argument for @computed
- not_computed context to disable resolver temporarily
- resolver signals
- enhancements:
- altered m2m handling with full through model expansion
- reduced transaction pressure in resolver
- new features:
-
0.2.10
- Fix related_query_name on M2M relations
- Fix computed fk fields to use instances and not the _id attribute (BREAKING CHANGE)
- Fix UNIONed resolver updates, that contain prefetches
- Fix thread isolation in signal handlers
-
0.2.9
- fix related_query_name issue
-
0.2.8
- Django 5.2 support
-
0.2.7
- setuptools issue fixed
-
0.2.6
- Django 5.1 support
-
0.2.5
- Django 5.0 & Python 3.12 support
- Django 3.2 support dropped
-
0.2.4
- performance improvement: use OR for simple multi dependency query construction
- performance improvement: better queryset narrowing for M2M lookups
ComputedFieldfor a more declarative code style added
-
0.2.3
- performance improvement: use UNION for multi dependency query construction
-
0.2.2
- Django 4.2 support
- Use
model._base_managerinstead ofmodel.objectsto prevent using overriddenmodels.objectswith a custom manager
-
0.2.1
- Django 4.1 support
-
0.2.0 - next beta release
-
new features:
- better memory control for the update resolver via
COMPUTEDFIELDS_QUERYSIZEor as argument on@computed - update optimization - early update-tree exit
- faster updates with
COMPUTEDFIELDS_FASTUPDATE checkdatacommandshowdependenciescommand- typing support for computed fields
- better memory control for the update resolver via
-
enhancements:
- better
updatedatacommand
- better
-
removed features:
- transitive reduction on intermodel graph (due to negative impact)
- pickled resolver map (due to showing low benefit)
update_dependent_multiandpreupdate_dependent_multi(due to showing low benefit and being a code nuisance)- Django 2.2 shims removed
-
bug fixes:
- regression on proxy models fixed
- sliced querset support for mysql fixed
-
-
0.1.7
- add list type support for
update_fieldsin signal handlers
- add list type support for
-
0.1.6
-
maintenace version with CI test dependencies changes:
- removed Python 3.6
- removed Django 2.2
- added Python 3.10
- added Django 4.0
- move dev environment to Python 3.10 and Django 3.2
Note that Django 2.2 will keep working until real incompatible code changes occur. This may happen by any later release, thus treat 0.1.6 as last compatible version.
-
-
0.1.5
- fix error on model instance cloning
-
0.1.4
- Django 3.2 support
-
0.1.3
- better multi table inheritance support and test cases
- explicit docs for multi table inheritance
-
0.1.2
- bugfix: o2o reverse name access
- add docs about model inheritance support
-
0.1.1
- bugfix: add missing migration
-
0.1.0
- fix recursion on empty queryset
- dependency expansion on M2M fields
m2m_changedhandler with filtering on m2m fields- remove custom metaclass, introducing Resolver class
- new decorator
@precomputedfor custom save methods - old depends syntax removed
- docs update
-
0.0.23:
- Bugfix: Fixing leaking computed fields in model inheritance.
-
0.0.22:
- Automatic dependency expansion on reverse relations.
- Example documentation.
-
0.0.21:
- Bugfix: Fixing undefined _batchsize for pickled map usage.
-
0.0.20
- Use
bulk_updatefor computed field updates. - Allow custom update optimizations with select_related and prefetch_related.
- Respect computed field MRO in
compute. - Allow updates on local computed fields from
update_dependentsimplifying bulk actions onComputedFieldsModel.
- Use
-
0.0.19
- Better graph expansion on relation paths with support for update_fields.
-
0.0.18
- New depends syntax deprecating the old one.
- MRO of local computed field methods implemented.
-
0.0.17
- Dropped Python 2.7 and Django 1.11 support.